Casting diversity in 2026: a data-driven look at modern cinema

Photo: Unsplash. The empty house tells you which audiences a studio expects, and which it does not.

For the past ten years CNTRST has tracked who appears on screen in wide-release theatrical films. The work is dull and repetitive: pull the credits, classify the cast, log it, do it again next month. Every spring we publish a state-of-the-data piece. This is the 2026 edition.

The dataset for this piece covers 412 films, released in the United States, the United Kingdom, the Netherlands, Belgium, France, and Germany between January 2024 and December 2025. We took the first seven billed actors of each film and recorded a publicly verifiable demographic profile for each. Where a studio publishes inclusion riders or behind-the-camera statistics, we logged those too. The full dataset is available on request.

The headline number is moving — slowly

Across the 412-film sample, the average share of non-white actors in the top-seven billed cast is 33.4%. That is up from 28.1% in our 2020 snapshot and 21.8% in our 2016 baseline. The trend line is real and it has not reversed.

It is, however, deeply uneven. The share for major-studio tentpole productions — defined as a production budget above $80 million — sits at 38.2%. The share for what we call "mid-budget originals" (between $15 million and $80 million) is 28.7%. The share for low-budget films released by major distributors is 26.4%. The pattern repeats year over year: the bigger the budget, the more visible the diverse casting becomes.

That is not a coincidence. Tentpole productions face the most public scrutiny and the most aggressive marketing spend. They are also the films most likely to be released into multiple international markets, where casting choices function partly as commercial decisions. Mid-budget originals are subject to less pressure and tend to draw their casts from a narrower pool of repeat collaborators.

Where the writers' rooms are

The on-screen number is the headline. The behind-the-camera number is more telling. Of the 412 films in the sample, 156 had publicly available writer credits with at least partial demographic data. Across those 156:

• 78% had a writers' room that was at least 50% white-identifying
• 14% had a writers' room with no women credited
• 22% had a sole credited writer (down from 34% in 2020)

The decline in sole-writer credits is the only number on this list that is moving in a direction many advocates would call positive. Sole-writer credits are increasingly being replaced by writers' rooms, which mechanically increases the chance that demographic representation among writers also improves. Whether the resulting films are better is a question CNTRST is not equipped to answer.

The streaming originals are pulling ahead

We restrict our theatrical sample to films that received a wide cinema release. A separate dataset tracks streaming originals — features and limited series produced directly for Netflix, Apple TV+, Amazon Prime, HBO Max, Disney+, and the major European public broadcasters. The streaming-original average non-white casting share for 2024–2025 is 41.2%.

That figure is consistently five to eight percentage points ahead of the theatrical average. Streaming platforms are not subject to the same multi-market release-window pressures as theatrical films, and several of them have explicit internal targets that we cannot independently verify but that visibly correlate with their published casting outcomes.

We will publish a more detailed breakdown of the streaming dataset later this spring. If you want to be reminded when it goes live, the contact address on our about page will forward you a one-line note. We do not have a newsletter.

What this number does not tell you

A few caveats worth repeating every year:

• Casting demographics are a single, blunt proxy for representation. A film can have a perfectly proportionate cast and tell a thoroughly stereotyped story.
• The dataset relies on publicly verifiable self-identification or, where unavailable, on the most conservative external classification we can find. Both methods have known limits.
• The sample skews heavily Anglophone. European-language productions are under-represented. We are working on it.

For a longer treatment of the methodology, see the about page. For the next CNTRST piece — which looks at the five 2025–2026 streaming series we think actually got representation right — see the streaming originals roundup.