The IMDb Top 100 re-ranked

Exaggerated reviews skew the numbers on most review platforms. What happens to the IMDb Top 100 if we remove the extreme ratings - the 1s and 10s? Full table below, but first, some trends.

Genre is highly correlated

The degree of reordering depends heavily on the genre. Family films (e.g. Disney), Animation, Sci-fi, Fantasy, and Romance move down the most. Maybe emotionally-resonant films get more 10s than they should? History, Thriller, Mystery, Crime, and Biography move up the most.

10s have a bigger impact than 1s

How far a film moves tracks its share of 10-star votes at −0.86 (a strong correlation). Its share of 1-star votes comes out at −0.50 because there aren't enough 1-star votes to move the needle. On almost every film in the Top 250, 1s are under 2.4% of all votes while 10s run anywhere from 20% to 36%. When a film falls in ranking, it's because the 10s were holding it up.

Full table

• # — where IMDb ranks it today.
• Stated — the score IMDb displays.
• Computed — the computed average from IMDb's own published histogram
• Adjusted — the same average with all 1s and 10s removed, then renormalized back onto a 1–10 scale so it's comparable to the columns beside it
• Δ Rank — how many places a film moves between IMDb's ranking and a ranking by Adjusted. Green with an up arrow means it moves up the list, toward #1. Red with a down arrow means it moves down.

All 250 films as of September 2026. Vote counts and histograms are IMDb's own published numbers.

IMDb's own adjustments

IMDb does its own adjustments from what the computed score would be, generally to weight credible reviewers more highly. In their words:

IMDb publishes weighted vote averages rather than raw data averages. Although we accept and consider all votes received by users, not all votes have the same impact (or 'weight') on the final rating. When unusual voting activity is detected, an alternate weighting calculation may be applied in order to preserve the reliability of our system. To ensure that our rating mechanism remains effective, we do not disclose the exact method used to generate the rating.

You can see it in the table above. Add up any film's published histogram, divide by the vote count, and you won't get the number IMDb shows you. The histogram sums to exactly the advertised vote total on all 250 films - no votes are thrown away - and yet the displayed score differs on 144 of them by more than a rounding error.

Trend analysis on their weights says that old movies are weighted higher and new ones lower. Films from the 1920s are scored about 0.16 points higher than their raw average. Films from the 2020s are scored 0.14 points lower, stepping down decade by decade almost without exception.

Release year is the single best predictor of the adjustment - knowing only what year a film came out gets you about a third of the way there. They also seem to discount films with a bunch of 10s. The rest is dictated by their own proprietary assessment of voter credibility.

The six biggest movers

Combining our re-ranking and IMDb's proprietary adjustment, here are the six biggest changes in film ranking.

Come and See ties Citizen Kane at 125 places down, so the bottom three is really a bottom four

The pattern in the histogram data across all six is that films that move up are the ones with a strong middle that IMDb's weighting had already punished for its 10s. The films that move down are the ones where the 10s dominated - whether due to a mobilized fanbase or classic reputation.