Roblox Games Losing the Most Players
Ranked by the share of its own player base each game shed over the last 6.7 hours, among the 838 games holding at least 500 players at both ends.
There is not yet a full day of history, so this chart measures the longest window the readings actually support, named by the true median age of the baselines used — 6.7 hours — rather than rounded to a day. On a window this short the daily cycle dominates almost everything: at the time of writing the median qualifying game is up sharply simply because the afternoon arrived, which means a game merely holding steady appears here as a heavy faller. Read the “vs tide” column rather than the raw percentage. The window extends itself toward 24 hours as collection continues.
- 1
32,889 → 2,819-92.4% vs tide-91.4%
- 2
2,603 → 804-70.1% vs tide-69.1%
- 3
7,614 → 2,819-63.9% vs tide-63.0%
- 4
5,216 → 1,942-63.7% vs tide-62.8%
- 5
3,333 → 1,288-62.3% vs tide-61.4%
- 6
1,414 → 573-60.4% vs tide-59.5%
- 7
1,520 → 616-60.4% vs tide-59.5%
- 8
3,387 → 1,447-58.2% vs tide-57.3%
- 9
1,178 → 527-56.2% vs tide-55.3%
- 10
4,386 → 2,000-55.4% vs tide-54.4%
- 11
2,756 → 1,257-55.4% vs tide-54.4%
- 12
39,396 → 18,090-55.0% vs tide-54.1%
- 13
1,657 → 791-53.2% vs tide-52.3%
- 14
1,397 → 677-52.5% vs tide-51.5%
- 15
3,598 → 1,802-50.9% vs tide-49.9%
- 16
19,144 → 9,599-50.8% vs tide-49.9%
- 17
1,656 → 850-49.6% vs tide-48.7%
- 18
1,086 → 581-47.5% vs tide-46.5%
- 19
4,628 → 2,483-47.3% vs tide-46.3%
- 20
17,736 → 9,558-47.1% vs tide-46.1%
- 21
9,095 → 5,006-45.9% vs tide-45.0%
- 22
939 → 517-45.9% vs tide-44.9%
- 23
1,859 → 1,047-44.6% vs tide-43.7%
- 24
1,086 → 613-44.5% vs tide-43.6%
- 25
1,145 → 650-44.2% vs tide-43.2%
- 26
9,197 → 5,320-43.1% vs tide-42.2%
- 27
1,457 → 853-42.4% vs tide-41.5%
- 28
87,144 → 51,315-42.1% vs tide-41.1%
- 29
2,657 → 1,566-42.0% vs tide-41.1%
- 30
11,654 → 6,900-41.8% vs tide-40.8%
- 31
3,244 → 1,923-41.7% vs tide-40.7%
- 32
1,796 → 1,085-40.6% vs tide-39.6%
- 33
3,327 → 2,017-40.3% vs tide-39.4%
- 34
12,566 → 7,635-40.2% vs tide-39.2%
- 35
23,955 → 14,675-39.7% vs tide-38.7%
- 36
3,424 → 2,099-39.7% vs tide-38.7%
- 37
2,429 → 1,513-38.7% vs tide-37.7%
- 38
87,536 → 54,556-38.6% vs tide-37.7%
- 39
809 → 506-38.4% vs tide-37.5%
- 40
3,467 → 2,213-37.1% vs tide-36.2%
- 41
68,744 → 44,522-36.2% vs tide-35.2%
- 42
926 → 604-35.7% vs tide-34.8%
- 43
1,495 → 977-35.6% vs tide-34.6%
- 44
10,528 → 6,881-35.6% vs tide-34.6%
- 45
1,297 → 851-35.4% vs tide-34.4%
- 46
851 → 565-34.6% vs tide-33.6%
- 47
993 → 661-34.4% vs tide-33.4%
- 48
4,570 → 3,053-34.2% vs tide-33.2%
- 49
891 → 596-34.1% vs tide-33.1%
- 50
6,114 → 4,100-33.9% vs tide-32.9%
Hourly player-count collection began , which is 10.3 hours ago across 5 completed runs. We have no readings from before that date and do not estimate any, so nothing here describes a period we did not measure. Nothing here should be read as evidence of long-term decline; that needs weeks of history and this dataset is hours old.
The difference between a dip and a decline
This is the chart most likely to be misread, so it is worth being blunt about what a big negative number here does and does not mean.
A dip is the daily cycle. Every game on Roblox loses most of its players overnight and gets them back the next afternoon. Read at the wrong hour, a thriving game posts a catastrophic-looking percentage. Over the short windows this chart currently uses, the great majority of what you see is this and nothing else.
A decay is the tail of a spike. A game that briefly tripled is giving those extra players back, and the percentages during that unwind are large and entirely expected. The important question is not how fast it is falling but where it settles — usually somewhat above where it started, which is a success.
A decline is a sustained downward trend across weeks, and it is the only one that means a game is in trouble. It is also the one this chart cannot see. Real decline is slow: a few percent a week, invisible in any single window, obvious in a three-month line. Detecting it needs the kind of history we simply do not have yet, and no amount of arithmetic on hours of data will produce it. When a game with a long record starts breaking below its own tracked range repeatedly, that will be worth showing — and it will need months behind it before it means anything.
How a missing reading could lie to you
The most dangerous artefact on a chart like this is not a wrong number. It is a missing one.
Suppose a game is read at 14:00, the 15:00 run fails for it, and it is read again at 16:00. A system that silently treats the 14:00 figure as the previous reading now compares two points two hours apart while believing they are one hour apart, and any movement between them is doubled in apparent rate. Worse, a system that fills the missing hour with a zero — which is the obvious lazy implementation — invents a total collapse and a total recovery that never happened.
Our collection writes a row only when Roblox actually answered. A failure goes into a separate table with its reason, never into the readings as a zero. Every comparison on this page then carries the true age of the baseline it used, which is shown in the window summary above and on each row, so a two-hour gap is visible as a two-hour gap rather than being quietly relabelled. The player-count charts on individual game pages take the same approach visually: the line breaks where readings are missing instead of drawing a straight segment across the gap.
None of that makes the data perfect. If Roblox itself reports a suppressed count during a problem on its side, we will store that number in good faith and it will appear here as a drop. That failure mode is real, we cannot currently exclude it, and the honest response is to say so rather than to imply a precision the pipeline does not have.
Questions about these charts
Almost never on its own. Over a short window the overwhelming cause of a decline is the clock — Roblox empties overnight and refills through the afternoon, and a game read on the downslope will post a large negative number while being in perfect health. The "vs tide" column is the one to read: a game down 12% on an evening when the median game is down 11% has done nothing unusual. A game down 40% while the median is down 5% has.
Because most spikes decay. A game that gained a wave of players from a video or a limited-time event gives most of them back once the reason ends, and a decay from an elevated peak produces large negative percentages for several days. That pattern — sharp rise, slower fall, settling above where it started — is the single most common story on Roblox, and it is not a failure. It is what a successful event looks like on the way down.
The same reason as on the gains chart, and the arithmetic is symmetrical. Small counts are dominated by random variation of roughly the square root of the count, so a 20-player game routinely swings 22% with nothing happening and would fill this chart with meaningless entries. Requiring 500 players at both ends of the window puts that jitter near 4.5%, and requiring it at the baseline end specifically means a game cannot appear here merely for having been tiny to begin with.
Not yet, and it is important to say so. A genuine decline is a downward trend sustained across weeks, visible only when you can compare this week against last month. What this page measures is a single window, which catches sudden events and nothing else. A game quietly bleeding two percent a week — the way games actually die — will never appear here at all, and would need a much longer baseline than any reading we currently hold.
A missed collection run, which turns two hours of gradual movement into what looks like one abrupt step. A Roblox-side error that returned suppressed counts for the current reading, which fabricates a collapse that reverses an hour later. And a game restarting its servers during our sweep, which briefly empties them. We record failed requests separately rather than writing them in as zeroes, so an outright failure never appears as a crash — but a low number that Roblox reported confidently is one we have no way to second-guess.
Two. If you play something on this list, a sharp unexplained fall relative to the tide is worth knowing about, because it usually means an unpopular update rather than a quiet afternoon. And if you build on Roblox, the decay profile of other games after their spikes is the most useful public information there is about how long attention lasts on the platform.
More Roblox charts
All Roblox charts → · Most played right now → · Trending games → · Biggest player gains →
Best Roblox games by genre → · Live Roblox player stats → · The full game database → · Working game codes →