Four Footnotes on Film Runtime: Follow-ups on Follows’s Latest Post
More answers to the question "Are two hour movies the new normal?"
Stephen Follows recently published an analysis of how movie runtimes have changed over the past four decades. I loved this article, and it gave me some follow-up questions I thought were worth answering.
Are adaptations longer than originals?
Do longer films stay in theaters longer?
Is every sequel longer than the last?
Do critics and audiences disagree more on long films?
Adaptations are longer than originals
Follows identifies budget and franchise status as the main drivers of runtime inflation. I want to drill down a little deeper and examine specific genres of source material that the film may have come from.
Novel adaptations and comic book films are the longest wide releases in every decade, averaging 121–124 minutes. Original screenplays are the shortest, at 106 in the 2000s and 112 today. The most dramatic inflation belongs to TV-series adaptations, which jumped from 100 minutes in the 2000s to 119 in the 2020s, possibly reflecting the migration of prestige TV properties to theatrical release (e.g., Downton Abbey: A New Era at 124 minutes, Downton Abbey: The Grand Finale at 123).
This makes intuitive sense. A novelist’s plot doesn’t compress neatly into 95 minutes and the fidelity pressure from a built-in fanbase makes cutting harder. An original screenplay, on the other hand, is written with a target runtime in mind. The inflation in original films (+6 minutes) is modest compared to adapted material.
Longer films hold the screen longer
Follows notes that audiences “keep showing up” for long films, and the box office backs him up. But showing up opening weekend is one thing. Coming back the next weekend is another. Knowing that longer movies are more of a commitment in theaters (the ticket is the same price, but the babysitter is hourly), I wouldn’t have been surprised at all if longer films had worse legs.
They don’t. Longer films genuinely have better legs.
The scatterplot above shows legs ratios for 3,310 wide releases earning at least $1 million. Legs improve modestly from about 2.8x for 90-minute films to 3.3x around the 130-140-minute mark, then flatten and even dip slightly, although sample size attenuates massively as we veer beyond the two-and-a-half hour zone. The overall correlation is statistically significant but the effect size is small; runtime explains only a fraction of the variance in legs. Pairwise tests confirm that the gains concentrate between 90 and 135 minutes; below 90 and above 135, the differences are indistinguishable from noise. In short, A two-hour film holds marginally better than a 90-minute one, but a 160-minute film doesn’t have better legs than a 130-minute one. Whether the mid-range benefit reflects quality selection, audience word-of-mouth dynamics, or theater scheduling incentives, the data doesn’t say.
Franchise escalation happens, but not always
Follows describes a “pattern of escalation” where each franchise installment runs longer than the last. He cites Star Wars, Indiana Jones, James Bond, and Fast & Furious as examples. Does this hold across all major franchises?
Not quite. It’s a strong tendency, but with notable exceptions.
Mission: Impossible is the clearest case. The franchise started at 123 minutes and has steadily climbed to 169 across seven wide releases. Fast & Furious (+34), James Bond (+35), the Conjuring Universe (+36), and the Saw franchise (+25) all also grow. But Transformers went the other direction, shrinking from 144 minutes down to 104 as the franchise pivoted from Michael Bay epics to the leaner Bumblebee, Rise of the Beasts, and the animated Transformers One. Harry Potter peaked with the middle installments and got shorter toward the end. The MCU, despite 37 entries, shows almost no net change. The first entry ran 126 minutes and the latest runs 127, even though individual films range from 105 to 181. The franchise as a whole never systematically escalated.
Critics and audiences agree more on long films
One version of the runtime-inflation story is that directors with leverage are self-indulgently padding their films. If that were true, you’d expect audiences to punish the relatively self-indulgent, long films relative to critics, or at least for the two groups to diverge. They don’t. They converge.
Both Rotten Tomatoes critic scores and audience scores rise monotonically with runtime. The median Tomatometer for sub-90-minute wide releases is 34%; for films over 150 minutes, it’s 77%. Audience scores follow the same curve, from 49% to 82%. But the gap between the two narrows. For short films, critics trail audiences by 15 points. For long films, the gap shrinks to 5.
This is partly a selection effect, as the films that get long theatrical runtimes tend to be the ones with the budget, talent, and studio confidence that also produce high scores. But the convergence is the interesting part. Critics are often accused of rewarding prestige bloat that regular audiences don’t enjoy. The data says otherwise. If anything, the films where critics and audiences agree most are the long ones.
Takeaways
Source material shapes runtime more than you’d expect. Novel adaptations average 121-124 minutes regardless of decade; comic book films range from 116 to 124, trending longer over time. Original screenplays are consistently the shortest wide releases, averaging 112 minutes today.
Longer films have better legs, but only up to a point. The relationship between runtime and holdover plateaus around 130-140 minutes. Beyond that, additional runtime doesn’t buy additional staying power. Not that increasing runtime actually buys any staying power; I’m just describing the relationship visible in the data.
Most major franchises have gotten longer over time, but escalation depends on creative continuity. Mission Impossible added 46 minutes across seven films; the Conjuring Universe added 36. But some franchises have static or shrinking runtimes as well, such as Transformers or the MCU.
Critics and audiences converge on long films, not diverge. The critic-audience gap on Rotten Tomatoes narrows from 15 points for sub-90-minute films to 5 points for films over 150 minutes. Whatever is driving runtime inflation, it also relates to audience and critic consensus.
Notes
Follows’ analysis draws on a different underlying dataset with different inclusion criteria, time windows, and category definitions than the CPRF Movie Database used here. I did not attempt to replicate his numbers exactly in this article, but I did try to replicate his findings to see if our databases were more or less aligned, and, consequently, my article could reasonably extend his findings. The findings were more or less replicated. Runtimes are trending upward, franchise films are longer than originals, and audiences keep showing up for long movies. The convergence of two independent datasets on the same broad conclusions suggests the patterns are robust, even where the specific facets of the underlying dataset differ.