CS2 aim stats: head share vs hit rate for 157 pro riflers

Bottom line: Per hit, a head hit does more: with the AK-47, 90% of head hits were the killing blow against 18% of chest or stomach hits, and with the M4A1-S 67% against 20% (every tier). Across 157 S and A tier pro riflers, though, the share of a rifler's hits that land on the head goes with only slightly better results (r = +0.21 with K/D, +0.16 with our own plain composite score), and most of the spread between riflers in kills per rifle shot, 83%, is how often shots hit at all. These records show where bullets landed, not where the crosshair was, so they cannot say which a player should aim at. Every figure is computed from the same match data behind the EsportsOdds CS2 data API as of 6 October 2026.

Terms used below:

• Killing blow: the hit that took the target's last health. Head share: head hits as a share of a player's bullet hits, from every gun.
• Hit rate: rifle hits as a share of rifle shots (AK-47, M4A1-S, M4A4, Galil AR and FAMAS). It counts every miss, including shots into smoke, so it is not a measure of skill.
• Gun hits and bullet hits: gun hits are a player's hits counted by weapon; bullet hits are the hits the body-part record counts, with grenade, fire and other non-bullet damage excluded. The two records agree closely in total, not match by match.
• Rifler: a player with at least 16,000 rifle shots and the AWP under 5% of their gun hits (157 players). S and A tier are the two highest grades of event prestige; B and C come below them.
• K/D is kills per death, ADR damage per round and KAST the share of rounds with a kill, an assist, a survival or a trade. The score is our own plain summary of the three: each divided by the median of the 157 riflers, then averaged, so 1.00 is the median rifler.

A head hit is the killing blow about three to five times as often as a chest or stomach hit

Start with what one hit does. For each rifle we took the player-sides (one player, one match, one side) in which that rifle made at least 90% of a player's gun hits, and at least 15 gun hits, looked at where those hits landed and counted how often the hit was the one that killed. These rows come from every tier, not only S and A, for sample size. An AK-47 head hit was the killing blow 90.0% of the time (89.5% to 90.5%) and a chest or stomach hit 18.1% (17.8% to 18.4%): 5.0 times as often. The M4A1-S figures are 67.0% and 20.4% (3.3 times as often) and the M4A4's 68.9% and 18.6% (3.7 times).

A head hit also does the most damage: 71 points on average with the AK-47, against 25 for a chest or stomach hit, and 62 and 63 with the two M4s, which fits their lower rate of killing blows. This is the game's damage rules showing up in the records, which any Counter-Strike player will expect; what the records add is how large the gap is in real rounds. Of the other parts the stomach is the most lethal: 23.9 killing blows per 100 AK-47 hits, against 16.3 for the chest, 14.3 for the arms and 16.4 for the legs.

Four cautions:

• Rifle and side cannot be pulled apart. The AK-47 is a T-side rifle and the M4s are CT-side rifles (99.9% of the AK-47 hits in these rows are on the T side, and none of the M4 hits). The M4A4 also has only 53 players and 2,227 hits.
• A row is mostly one rifle, not only one. Up to a tenth of a row's hits can come from other guns, and the data classes the AWP with the rifles. Keeping only rows in which the rifle made at least 99% of the gun hits leaves 151 AK-47 players (4,439 hits), who give 92% against 18% (5.0 times as often, 4.7 to 5.4); the M4A1-S gives 61% against 21% (2.8 times, 2.2 to 3.6, from 29 players) and the M4A4 68% against 19% (3.6 times, 1.9 to 5.8, from 5 players). Treat the M4 figures as about three times.
• A killing blow depends on what came before. The target may already have taken teammates' damage or earlier bullets in a spray, so a chest hit that lands third in a spray can be the killing blow while a first-bullet head hit on a full-health target in a helmet may not be. The figures describe hits as they landed in real rounds.
• The kills in a hit group are read as the kills whose last hit landed there; the check is in the methods.

Most of the spread between riflers is how often shots hit

A head hit does more than a body hit, but a rifler still has to land it. A rifler's kills per rifle shot is the hit rate times the kills each hit brings, and head share lives in the second part. On a log scale the two add, and across the 157 riflers hit rate accounts for 83% of the spread in kills per rifle shot (95% interval 78% to 89%) and kills per hit for the rest. The split is much the same with each rifle:

That is a statement about how spread out these riflers are on each part, not about what a point of each does. In plain terms: a rifler one standard deviation above the rest in hit rate (2.4 points) gets about 16% more kills per rifle shot, and one standard deviation above in head share (2.9 points) about 5% more, taking the across-rifler slope of kills per hit on head share (1.8% per point). Counted a second way, by each measure's squared correlation with kills per rifle shot across the 157, hit rate accounts for about 88% and head share for about 12%; the two are almost unrelated to each other, so the shares add to about 100%, and both ways of counting say most of the spread is hit rate.

Kills per hit follows head share closely across riflers (r = +0.89, about four-fifths of the variation), partly by construction, because a head hit is far more often the killing blow. The AK-47 gets 5.17 kills per 100 shots and 59.0% of them are headshots, against 40.7% for the M4A1-S and 46.6% for the M4A4, which fits the AK-47's higher rate of killing blows on head hits in the section above.

Do riflers with more head hits score better?

Only a little, and about as much as hitting more often does. To compare results we need a measure of how a rifler did. The match records carry no standard performance rating, and the per-match rating our API serves is our own early formula, kills plus half the assists over deaths, which adds little to K/D. So the score used here is our own and deliberately plain: a rifler's K/D, ADR and KAST, each divided by the median of the 157, then averaged. It is not a published rating and leaves out opening duels, clutches and everything else a rating might weigh. It runs +0.98 with K/D and +0.91 with ADR, so it is mostly those two, and K/D and ADR depend on a rifler's role, team and the opposition faced. It is a rough, transparent summary of how a rifler's rounds went.

Across the 157, this is how each result lines up with head share and with hit rate (correlation, with its 95% interval):

Of the scoreboard results, K/D is where head share shows most, and its interval stays above zero; hit rate's interval for K/D reaches zero. For ADR neither measure shows a link, and for KAST only hit rate does. Put both measures into one model and each keeps about its own size (for K/D, +0.21 for head share and +0.14 for hit rate, in standard deviations of K/D per standard deviation of the measure). Leaving out any one of the 157 moves head share's correlation with the score only between +0.12 and +0.19.

Part of the link with K/D is built in. A headshot kill is a head hit, so it is counted in both the K/D and the head share, and head share runs +0.93 with the share of a rifler's kills that ended on a head hit. Take those kills out of K/D and the link with head share turns clearly negative (−0.61), because riflers with a higher head share get more of their kills from head hits and fewer from body hits, with total kills little changed; taking head-hit damage out of ADR does the same (−0.67). So the +0.21 shows that a high head share does not come at a cost in K/D, not that it adds to it.

Role is a second reason for care. Riflers who take more opening duels (a first kill or a first death in a round) deal more damage per round (r = +0.54) and put slightly more of their hits on the head (r = +0.18). Adding opening-duel share to the two-measure model leaves head share's coefficient for K/D at +0.19 (from +0.21) and cuts its coefficient for ADR from +0.12 to +0.02.

Now split the riflers at the medians. Across the 157, hit rate and head share are not related (r = +0.05, 95% CI −0.11 to +0.20; ten other cuts agree, in the methods), so the four groups are the four corners of a round cloud. "Hit more" means a hit rate above the median (16.5%) and "head more" a head share above the median (21.9%); a rifler exactly on a median goes in the lower group.

The 36 riflers who hit more and put fewer hits on the head average a score of 1.005 and the 36 who hit less and put more on the head 1.005: a gap of under 0.001, against a spread of 0.062 (one standard deviation) across the 157. With 36 riflers in each group the 95% interval (−0.026 to +0.026, about 0.4 of that spread) cannot rule out a difference of that size, so read it as no detectable difference, not as proof that the two profiles are equal. Riflers who hit more and put more on the head score 0.022 higher on average than those who hit less and put less (−0.006 to +0.050). Take the 31 riflers with the highest head share and the 31 with the lowest: the top group's mean score is 0.033 above the bottom group's (95% interval −0.003 to +0.066); for hit rate the same comparison gives 0.020 (−0.004 to +0.048).

Two riflers at the ends of head share

The two ends of head share make a useful illustration. Of the 157 S and A tier riflers, zont1x puts the lowest share of hits on the head, 14.8%, against a median of 21.9%, and malbsMd the highest, 30.7%. Neither is a small-sample accident: zont1x's head share rests on 13,764 hits and malbsMd's on 11,811. Here are their other figures:

These are records of what happened in matches, not a verdict on either rifler's ability. To ask for a name or a figure to be corrected or removed, see our player data notice.

On the parts, the two differ in different directions. zont1x turns more rifle shots into kills (5.71 per 100 shots against 5.06; a difference of 0.65, 95% interval 0.35 to 0.93), with fewer headshots among those kills (35.1% against 64.0%), and has the higher KAST (4.7 points more, 3.0 to 6.5); malbsMd deals more damage per round (5.2 points more, 1.9 to 8.8); and K/D does not separate them (a difference of +0.05, −0.01 to +0.11). Which of the two looks better depends on how those parts are weighed, which is one reason the comparison above uses all 157 riflers.

The missing head hits show up in the stomach. Here are the two riflers' hits by body part, on one colour scale:

Against all 157 riflers pooled (21.9% head, 47.8% chest, 14.5% stomach), zont1x is 7.1 points light on head hits, and 5.3 points of that turns up in the stomach: 19.8% of the hits, the most of the 157, against 14.5%. The chest share, 48.0%, is about the same as the pool's. malbsMd is at the other end of the stomach (10.1%, the least of the 157) and also has the least chest (44.6%). The pattern holds across the 157: the more head hits, the fewer stomach hits. The records show where the hits landed, not why: where the crosshair sat is not in them.

The correlation is −0.89. Some of that is arithmetic: the shares have to add up to 100, and if extra head hits came out of every other part in proportion, the stomach would already correlate at about −0.80 with head share. What is not automatic is how the trade-off splits. Proportional crowding would cost the stomach 0.19 of a point and the chest 0.61 for each extra point of head share. In the data a point costs the stomach 0.59 of a point and the chest 0.15 (the arms 0.10 and the legs 0.17). The chest correlates with head share at only −0.32, where proportional crowding would give about −0.95, and it runs from 44.6% to 50.9% of hits; the stomach carries the trade.

The same rifler in a series with more head hits

Comparing riflers mixes in role, team and opponents. Comparing a rifler with themselves holds the rifler's role and usual style roughly fixed, though not the opponent, the map or the team. For each of 13,134 rifler-series (a rifler's whole series, a best-of-one to a best-of-five, with at least 40 bullet hits, 100 rifle shots and 15 rounds, from 157 riflers) we measured how far head share, hit rate and the result sat from that rifler's own average. A series' score is its K/D, ADR and KAST over the same 157 medians as before.

A series with a head share one standard deviation above the rifler's usual (4.9 points) went with 0.036 more kills minus deaths per round, and a series with a hit rate one standard deviation above usual (3.4 points) with 0.036 more: the same, though a point of hit rate moves it about 1.4 times as much as a point of head share, because the head-share standard deviation is wider. For damage per round the two differ, 2.1 points for head share and 3.7 for hit rate. In the result's own standard deviations:

Neither measure moved the rifler's chance of winning the series by more than about two percentage points (−0.3 points for head share and +0.9 for hit rate): the interval for head share (−1.2 to +0.5) includes zero, and the one for hit rate (0.0 to +1.8) only just reaches it.

Much of this is arithmetic rather than strategy. A headshot kill is one head hit and one kill, so a series with more head hits is a series with more kills by construction. At least 74% of a rifler's series-to-series variation in head share is what independent hits at their own average rate would produce (for hit rate, at least 29%; independent hits are a floor for chance, not a model of it), so a higher head share is mostly hits that happened to land high, and a head hit kills more often than a body hit. It describes what a better series looks like, not how to produce one.

What the records show

Three results stand out.

• Per hit, head hits do more. A head hit was the killing blow 3.3 to 5.0 times as often as a chest or stomach hit, with each of the three rifles.
• Per shot, how often shots land accounts for most of the spread between riflers. Hit rate accounts for 83% of the spread in kills per rifle shot; one standard deviation of hit rate goes with about 16% more kills per shot and one of head share with about 5%.
• Between professionals, no detectable difference between the profiles. Riflers who hit more and put fewer hits on the head and riflers who hit less and put more on the head average scores under 0.001 apart (−0.026 to +0.026), and riflers above the median on both measures score 0.022 higher than riflers at or below it on both (−0.006 to +0.050).

What the records cannot show is what a rifler should aim at. They say where bullets landed, not where the crosshair was. A low head share can mean aiming at the chest, or aiming at the head and landing lower; a high one can mean the opposite. The figures cannot tell these apart. They show how often a hit at each height was the killing blow and how professionals' results line up, not what any one of them would gain by aiming differently.

For related studies, see what the AWP nerf actually changed, how the weapon meta moved and which weapons decide rounds. The per-match weapon (shots and hits) and hit-group records behind this one are served by the API's match depth endpoint for most matches, so the same analysis can be run on your own cuts.

What this data cannot see

Everything here comes from per-match totals, which sets hard limits.

• No aim, only landing. The records say where bullets landed, not where the crosshair was or what a rifler was trying to hit, so head share is an outcome, not a choice, and nothing here shows what aiming at the head or at the body would do for any one rifler.
• The score is ours, and rough. It averages K/D, ADR and KAST, which are scoreboard totals for the whole match and every gun, while hit rate counts rifles only. It ignores opening duels, clutches, the economy, a rifler's role and the strength of the opponents faced and of the rifler's own team, so differences of a few hundredths are not meaningful.
• No distance, round or situation. A long-range spray, a point-blank tap and a shot into smoke all look alike in a match total.
• No pose and no crosshair. Whether the target was crouching or moving changes which part is exposed, and where the crosshair sat is not recorded.
• Everyone here is a professional. A pro sample is already selected for aim, and that kind of selection can hide or distort a link that exists among all players.
• Hit rate is not skill. It counts every miss, and a rifler's role, how many shots are fired and whether the rifler plays on the T or CT side all move it. The comparisons between named riflers describe their records, not their ability.
• Head share spans every gun. Hit rate and kills per hit cover rifles only, while head share includes pistols, SMGs and the AWP. A figure built from head hits over rifle shots, as the appendix allows, would mix hits from every gun with shots from rifles only.
• A series is a whole match. A best-of-three mixes maps, sides and opponents, and the within-rifler comparison holds none of them fixed.
• The body part is the game's bookkeeping. The recorded split between the arms and legs changed with a game update in October 2024, and the window starts after it. We take each recorded body part to be the victim's own, as in the diagram; nothing in our data lets us check that independently.

How we measured it

• Window. Completed matches with game version cs2, scheduled from 4 November 2024 to 30 September 2026 (UTC): 11,029 matches and 3,207 players with hit-group records. That is 82.2% of the 13,418 completed matches in the window, and 99.4% at S and A tier; the rest have no bullet hits recorded by body part. The window begins after the 28 October 2024 update, which changed how the game records hits on the arms and legs, and after the week that followed it, and stops before the current, still-filling month.
• Tiers. Tier is an event's prestige grade: S highest, then A, B and C. The killing-blow comparison uses every tier for sample size (11,029 matches); everything else uses S and A tier events (2,315 matches), apart from the B and C tier replication in the table of cuts below.
• Bullet hits. Body-part hits recorded for a player, from per-match hit-group records, with grenade, fire and other non-bullet damage excluded. The body part is the victim's. The records may include a sliver of damage to teammates, which this data cannot separate.
• Rifler. At least 16,000 rifle shots (the AK-47, M4A1-S, M4A4, Galil AR and FAMAS; the AUG and SG 553, under 0.2% of rifle-class shots, are left out) with the AWP under 5% of the player's gun hits: 157 players. The threshold counts shots fired, not hits, so it does not select on hit rate directly. Among the 157, though, shots fired and hit rate do correlate (−0.26 against the log of shots fired), which is why the table below also tests 12,000 and 24,000. AWPers are a different population with a very different hit rate, and they are left out of this study.
• Killing blows per hit. The comparison by rifle uses the player-sides (one player, one match, one side) in which the AK-47, M4A1-S or M4A4 made at least 90% of the player's gun hits, with at least 15 gun hits, in every tier (845 AK-47 players, 228 M4A1-S players and 53 M4A4 players have such rows); the 99% cut keeps rows in which the rifle made at least 99% of the gun hits. Hits, damage and kills come from the hit-group record of those player-sides. We read the kills in a hit group as the kills whose last hit landed there: in the S and A rows the head kills from the hit-group record total 367,931 against 366,394 headshots on the scoreboards, 0.4% apart, which fits. Chest or stomach pools those two parts. Intervals resample players and keep each player's rows together (2,000 resamples; seed 20261006). The ratio is the head figure over the chest-or-stomach figure, resampled the same way.
• Score and scoreboard figures. K/D, ADR and KAST come from the match scoreboard for the 157 riflers: 14,039 rifler-matches, the riflers' scoreboard rows for which the weapon and hit-group records also exist. A row's rounds are its damage divided by its ADR, which agrees with the rounds in the map score to within one round on 97.7% of all S and A tier scoreboard rows (98.1% of the riflers' rows). A rifler's K/D is kills over deaths, ADR total damage over rounds and KAST the rounds-weighted mean, all pooled over the window; the score is the mean of the three, each divided by the median of the 157 (K/D 0.97, ADR 72.7, KAST 72.4%). Opening-duel share is first kills plus first deaths per round; across the window those sum to between 0.195 and 0.202 per player-round in every month, where ten players and one opening duel a round would give 0.200.
• Split of kills per shot. A rifler's kills per rifle shot is hit rate times kills per hit. Taking logs makes the variance across riflers a sum of the two parts' variances and twice their covariance, and each part's share is its variance plus its covariance over the total, so the two shares add to 100%. The split uses the riflers with at least 3,000 shots with the rifle in question (157 for the AK-47, 155 for the M4A1-S and 98 for the M4A4; all 157 for the five rifles together), with an interval that resamples riflers (2,000 resamples). The plain-unit figures: one standard deviation of hit rate is a change of 0.147 in its log, which with kills per hit held fixed is that proportion more kills per shot; for head share it is the across-rifler slope of the log of kills per hit on head share times one standard deviation of head share. The shares of the variation are the squared correlations of each measure with kills per shot.
• Between riflers. Correlations are Pearson with Fisher-z 95% intervals; the rank correlations agree (head share with K/D +0.19 and with the score +0.14; hit rate with K/D +0.16 and with the score +0.13). The two-measure model standardises each variable and takes bootstrap intervals over riflers (4,000 resamples). The groups split at the medians, with the one rifler exactly on each median placed in the lower group; the fifths are 31 riflers each, and the intervals on group means and gaps resample riflers (2,000 resamples). The share of a rifler's kills that ended on a head hit is the hit-group head kills over the scoreboard kills. Removing head kills from K/D takes the kills whose last hit landed on the head out of the numerator; removing head-hit damage from ADR does the same for damage.
• Within a rifler. A rifler-series needs at least 40 bullet hits, 100 rifle shots and 15 rounds. A series' score is its kills over its deaths (deaths floored at 1, which never binds here), its damage over its rounds and its rounds-weighted KAST, each divided by the same 157 medians. Each measure and each result is the series' deviation from that rifler's own average over their series (a rifler fixed effect); the two deviations go into one regression per result, the slope is scaled by one standard deviation of the measure (as a deviation from the rifler's own average) and, for the chart, divided by the standard deviation of the result. Intervals resample riflers (2,000), which treats teammates and opponents in the same series as independent, so the interval on the series result is slightly too narrow. The picture holds with 80 or more bullet hits, 200 or more rifle shots, 25 or more rounds, or only series whose rounds agree with the map score: kills minus deaths per round moves by 0.17 to 0.21 standard deviations for each measure and the score by 0.15 to 0.21. The share of a measure's variation that independent hits would produce compares the variance of the series' deviations with the mean of p(1 - p) divided by the series' hits (or shots), at the rifler's own average rate p.
• The two named riflers. Their intervals resample their own matches (118 and 123; 4,000 resamples), holding the medians of the 157 fixed.
• Body parts. The part-by-part comparison uses plain correlations and least-squares slopes of each body part's share on head share across the 157; the crowding comparison divides a part's median share by the median non-head share (78.1%).
• Intervals. Correlation intervals are Fisher-z 95% intervals unless stated. The sampling errors for hit rate and head share in the appendix (0.4 and 0.5 points for a typical rifler) are the median half-width of each rifler's 90% match-level bootstrap interval (2,000 resamples, fixed seed) divided by 1.645; plain binomial errors, which treat every shot as independent, would be 0.2 and 0.4 points.
• Gaps in the records. In 54 of the 11,029 matches (all played 22 to 26 January 2026, from six events) the hit-group record is almost empty, and 7 pairs of matches carry identical player totals under different match IDs, so one match in each pair is probably counted twice. We left both in, because the API serves the same records.
• What was not pre-registered. None of these tests was. We looked at more cuts while building the study than appear here, so a result near a margin, such as the leave-one-out range or the lower end of an interval, should be read as exploratory. The comparisons that carry the claims above are far from any margin, and each one that bears on a claim is reported in the text or the tables.
• Rounding. Percentages are rounded once, from raw counts.

Is hit rate linked to head share? The correlation across the 157 riflers under eleven cuts, none pre-registered (the main cut came first, the others were added as checks on it, and the rank correlation last). No interval excludes zero and none lies wholly beyond ±0.15; the main interval reaches +0.20, so a small positive link cannot be ruled out, while a strong one can.

Rifle-dominant rows are player-sides in which the five rifles made up at least 90% of the player's gun hits (with at least 15 gun hits); that cut keeps the players with 300 or more bullet hits in such rows. The adjusted hit rate subtracts what the league's rate with each rifle predicts for the player's mix of shots. The replication on 660 B and C tier players in a separate set of matches (87 of them also appear in the main set, in different matches) gives −0.02 (−0.09 to +0.06), and −0.02 (−0.10 to +0.06) without those 87. Noise in the measurements is not hiding a link: split each player's matches into random halves, 1,000 times over, and on average the two halves' hit rates correlate at 0.95 and their head shares at 0.93 (0.97 and 0.96 corrected to the full sample); correcting the headline correlation for that noise leaves r = +0.05.

Counts come from per-match weapon and hit-group records in our dataset. See our methodology page for how our data is sourced.

Appendix: all 157 riflers

Alphabetical. Matches, rifle shots, rifle hits, head hits and bullet hits are over the window; hit rate is rifle hits per rifle shot, and head share is head hits per bullet hit, from every gun. Rifle shots here are counted over the player-sides that also have a hit-group record, so the 157 rows sum to 6,789,819, 7 shots below the 6,789,826 in the table of kills per rifle shot; no hit rate changes at one decimal. A typical rifler's hit rate and head share each carry a sampling error of about 0.4 and 0.5 points (see the methods). To ask for a name or a figure to be corrected or removed, see our player data notice.

Cite this

EsportsOdds, "CS2 aim stats: head share vs hit rate for 157 pro riflers", 5 October 2026. https://esportsodds.gg/learn/cs2-aim-stats

Copy citation