The anatomy of a produced screenplay.
What can 427 produced feature screenplays from 2000-2026 tell us about what is on the page? Explore five measurements, their variation and the evidence behind them.
Published 13 August 2026. Updated 4 November 2026. Measured with engine 1.6. Methodology
Page explanation updated 4 November 2026. Data release: 4 November 2026.
Corrected 4 November 2026. Scene-heading detection has been corrected and measurements rerun on the same source files. An incomplete copy of Saw is excluded, and unresolved extraction issues are withheld from affected measurements. The corrected corpus contains 427 sources; eligible samples vary by measure.
Read the correction noteThe findings at a glance
Each figure is a separate median. Select one to see its context.
Jump to a section
How long a produced feature actually is.
The famous rule of thumb says a page a minute and a hundred and twenty pages. In this eligible sample, the middle half of script lengths spans 25 pages, with an upper quartile of 131.
View data table
| Interval | Scripts |
|---|---|
| 60 to <70 | 1 |
| 70 to <80 | 4 |
| 80 to <90 | 17 |
| 90 to <100 | 41 |
| 100 to <110 | 78 |
| 110 to <120 | 93 |
| 120 to <130 | 72 |
| 130 to <140 | 46 |
| 140 to <150 | 26 |
| 150 to <160 | 15 |
| 160 to <170 | 16 |
| 170 to <180 | 6 |
| 180 to <190 | 5 |
| 190 to <200 | 2 |
| 200 to <210 | 2 |
| 210 to <220 | 1 |
| 220 to <230 | 1 |
| 230 to <240 | 0 |
| 240 to <250 | 0 |
| 250 to <260 | 1 |
| Total eligible scripts | 427 |
median 117; eligible n=427. Intervals include the lower bound and exclude the upper bound. The final “or more” interval, where shown, includes the grouped tail.
Page count, n=427 eligible produced features 2000-2026. Snapshot anatomy-of-a-produced-screenplay-2026-11-v6
Use the range as context, not a target. Consider whether every page earns its place in the story you want to tell.
The package also retains a median modelled runtime of 117 minutes (n=379). This is an estimate from script composition, not observed film length. The model has not been validated on this full cohort; see the runtime assumptions and limits.
How scene length varies across scripts.
The median scene count is 124 (n=344). Scene length is a separate measure: we take each script’s median scene length, then the median across those scripts. That result is 4 eighths of a page (n=344). It does not mean half of all individual scenes have that length. For a breakdown by genre, read How long are screenplay scenes?
View data table
| Interval | Scripts |
|---|---|
| 1 to <2 | 13 |
| 2 to <3 | 64 |
| 3 to <4 | 86 |
| 4 to <5 | 54 |
| 5 to <6 | 60 |
| 6 to <7 | 25 |
| 7 to <8 | 14 |
| 8 to <9 | 6 |
| 9 to <10 | 6 |
| 10 to <11 | 3 |
| 11 to <12 | 5 |
| 12 or more | 8 |
| Total eligible scripts | 344 |
median 4; eligible n=344. Intervals include the lower bound and exclude the upper bound. The final “or more” interval, where shown, includes the grouped tail.
Median scene length per script, in eighths of a page; scripts above 12 eighths grouped at the right, n=344 eligible produced features 2000-2026. Snapshot anatomy-of-a-produced-screenplay-2026-11-v6
View data table
| Interval | Scripts |
|---|---|
| 20 to <40 | 6 |
| 40 to <60 | 13 |
| 60 to <80 | 26 |
| 80 to <100 | 54 |
| 100 to <120 | 56 |
| 120 to <140 | 51 |
| 140 to <160 | 45 |
| 160 to <180 | 30 |
| 180 to <200 | 20 |
| 200 to <220 | 12 |
| 220 to <240 | 9 |
| 240 to <260 | 6 |
| 260 to <280 | 4 |
| 280 to <300 | 4 |
| 300 to <320 | 3 |
| 320 or more | 5 |
| Total eligible scripts | 344 |
median 124; eligible n=344. Intervals include the lower bound and exclude the upper bound. The final “or more” interval, where shown, includes the grouped tail.
Scene count; scripts above 320 scenes grouped at the right, n=344 eligible produced features 2000-2026. Snapshot anatomy-of-a-produced-screenplay-2026-11-v6
60 places, mostly indoors, mostly daylight.
The median measured script has 60 detected locations. Scene-heading names are normalised to group recognised places; naming variants can still split or merge locations.
View data table
| Interval | Scripts |
|---|---|
| 0 to <15 | 4 |
| 15 to <30 | 32 |
| 30 to <45 | 50 |
| 45 to <60 | 83 |
| 60 to <75 | 73 |
| 75 to <90 | 36 |
| 90 to <105 | 21 |
| 105 to <120 | 22 |
| 120 to <135 | 10 |
| 135 to <150 | 6 |
| 150 to <165 | 1 |
| 165 or more | 6 |
| Total eligible scripts | 344 |
median 60; eligible n=344. Intervals include the lower bound and exclude the upper bound. The final “or more” interval, where shown, includes the grouped tail.
Unique locations; scripts above 165 locations grouped at the right, n=344 eligible produced features 2000-2026. Snapshot anatomy-of-a-produced-screenplay-2026-11-v6
62%
of scenes play interior on the median script
Middle half: 53% to 72%
n=344 eligible features
60%
of time-stamped scenes play in daylight
Middle half: 46% to 72%
n=304 eligible features
Big casts, concentrated voices.
42 characters speak in the median measured script. Separately, the median share of dialogue attributed to a script’s largest speaking part is 29.2%. “Speaking characters” counts detected character cues, not principal roles; extraction can occasionally split one name or misread a line as a cue.
View data table
| Interval | Scripts |
|---|---|
| 0 to <10 | 3 |
| 10 to <20 | 16 |
| 20 to <30 | 33 |
| 30 to <40 | 52 |
| 40 to <50 | 54 |
| 50 to <60 | 30 |
| 60 to <70 | 15 |
| 70 to <80 | 11 |
| 80 to <90 | 6 |
| 90 to <100 | 3 |
| 100 to <110 | 1 |
| 110 or more | 3 |
| Total eligible scripts | 227 |
median 42; eligible n=227. Intervals include the lower bound and exclude the upper bound. The final “or more” interval, where shown, includes the grouped tail.
Speaking characters; scripts above 110 characters grouped at the right, n=227 eligible produced features 2000-2026. Snapshot anatomy-of-a-produced-screenplay-2026-11-v6
29.2%
of all dialogue belongs to the largest part on the median script
Middle half: 23.1% to 35.4%
n=227 eligible features
1.14
lines of dialogue per line of action on the median script
Middle half: 0.78 to 1.61
n=379 eligible features
The shape of each genre.
The middle bands overlap on many measures. For length, the page-count comparison below identifies the longest and shortest medians. These are patterns in this sample, not requirements for a genre.
Compare one genre with the full sample
View data table
Scroll sideways to see all columns.
| Genre | Eligible n | Lower quartile | Median | Upper quartile |
|---|---|---|---|---|
| Action | 53 | 116 | 128 | 152 |
| Thriller | 30 | 114.25 | 126 | 138.25 |
| Sci-Fi | 32 | 109 | 121 | 139.5 |
| Romance | 25 | 103 | 117 | 130 |
| Drama | 105 | 103 | 115 | 138 |
| Animation/Family | 41 | 101 | 115 | 128 |
| Comedy | 66 | 107 | 114 | 120.75 |
| Horror | 45 | 100 | 110 | 118 |
| Fantasy | 17 | Not shown | Not shown | Not shown |
| Crime | 13 | Not shown | Not shown | Not shown |
The middle half lies between the lower and upper quartiles. Results below n=20 are not shown. These ranges describe this corpus, not a target or quality score.
n per genre shown beside each label. 2000-2026. Snapshot anatomy-of-a-produced-screenplay-2026-11-v6. Dot: median; band: middle 50%
Fantasy, Crime are measured but held back until they reach 20 eligible scripts for a measure: publishing a median over a handful of titles invites more confidence than the sample can carry. They unlock as the corpus grows.
What this means for a working writer.
The middle 50% is the range between the lower and upper quartiles. Half of this sample sits outside that band. Being inside it does not establish quality, and being outside it does not identify a problem. Use a comparison to frame a question about your draft, not as a target to write towards.
The sample is selective: available produced scripts often come from awards campaigns, and draft formats vary. The measured ranges for different genres overlap. These figures cannot tell you whether a particular scene works, how a reader will respond, or whether a script will be produced.
Methodology in brief
- Cohort: Produced feature screenplays released 2000-2026, measured from the file we hold, excluding parses the engine rated low confidence. 1 of 428 corpus scripts were excluded, each with a stated reason, listed in the snapshot manifest.
- Structural counts come from a deterministic measurement engine (v1.6). Runtime is a fitted estimate. Repeatability does not prove extraction accuracy. No AI judgement scores or user scripts are included.
- Figures are medians with interquartile ranges. Genre cells below n=20 are suppressed. Publicly available produced scripts skew toward acclaimed titles, so this cohort describes produced screenwriting as it survives publication, not a random sample of everything shot.
Explore your own screenplay
Upload your screenplay to see its page, scene, location and speaking-part counts. The measurements and their export are available on the free plan, without an AI analysis. Use this research as context, not a score your draft needs to match.
Measuring requires a free account and a laptop or desktop. On a phone, we’ll email you a link for your computer. Plan storage limits apply.
Publication history
Reviewed release Current release
Scene-detection correction using method 1.6 on the same 428 v5 source PDFs. Excludes the incomplete Saw PDF (427 retained); retains previous holds and withholds unresolved image content and heading-count discrepancies. Eligible samples: pages 427, scenes and locations 344, dialogue 379, characters 227. Prior releases remain available unchanged.
Source review of all 448 in-range held files corrects draft writing credits and public provenance. Nine previously included files are removed after source-identity or screenplay-extent checks, leaving 428 features. Shared cue-quality checks and 68 targeted source-page reviews withhold suspect measurement families; the No One Will Save You exception is withdrawn. Median PDF pages is 117 (n=428), scenes 124 (n=401), locations 62 (n=401), and speaking characters 42 (n=233). Page counts include cover and front matter. This is a targeted review, not a whole-corpus accuracy estimate. Prior releases remain unchanged.
437 features; Nell excluded, Memory identified as the 2023 drama, and figures use eligible samples. Median 117.5 pages (n=436).
Introduced measurement-family exclusions. Nell and Memory metadata were later corrected in v4; original histograms did not apply these exclusions.
Expanded to 438 scripts. This version predates measurement-family guards.
268 produced feature scripts.
Cite and reuse this research
Share a finding, check the evidence or download the data behind this release.
The PreList Industry Intelligence (2026). "The Anatomy of a Produced Screenplay." Cohort: 427 produced feature screenplays, 2000-2026; eligible n varies by measure. Snapshot anatomy-of-a-produced-screenplay-2026-11-v6. https://www.theprelist.io/research/anatomy-of-a-produced-screenplay (version: https://www.theprelist.io/research/versions/anatomy-of-a-produced-screenplay-2026-11-v6)
Licensed CC BY 4.0: reuse the figures, charts and dataset with attribution and a link back to this page. Snapshot anatomy-of-a-produced-screenplay-2026-11-v6; every number above traces to it. Screenplay text and third-party imagery are not included in this licence.
