How Uneven Does a Chore Wheel Get in a Month?

The short answer

The fix for an uneven month is a setting, not more spinning: random spins for one-off jobs, Remove the winner after each spin for anything that repeats, and a fixed rotation when it matters who gets which job.

• One spin a day, winner left on the wheel. With 4 people over 30 days the fair share is 7.5 turns each, but the busiest person ends 7 or more turns ahead of the luckiest, a week's worth, in 33.5% of months, and the same name comes up 3 days running in 77.9%.
• Longer doesn't fix it. For 4 people the expected gap grows from 5.6 turns after a month to 19.7 after a year.
• Remove the winner until everyone has had a turn. Counts never differ by more than 1, and no one gets it 3 days running. One repeat remains: the person who ends a round starts the next 1 time in 4, so a 4-person month has at least one back-to-back 86.7% of the time.
• Several chores a day. Spin each of 3 chores for 3 kids and some kid gets two or more on 77.8% of days. Dealing them gives one each, but who gets the dishes evens out only with a fixed rotation.
• An older kid weighted double (Older *2) gets 2/3 of the turns on average with one sibling, but in 33.3% of months their share is 10 or more percentage points off.

These figures are exact, worked out by counting every possible month, and checked against the simulation described at the end. To try your own household, use the simulator further down.

One spin a day: how uneven a month gets

The setup is the names version of the chore wheel: the names go on the wheel, you read out a chore, and you spin for who does it. Remove the winner after each spin is off by default there and on the random name picker, so every spin gives each of 4 people the same 1-in-4 chance, whatever happened yesterday. Kids or roommates, the numbers are the same.

The figures count spins, not dates. If you spin once a week, as the chore wheel page suggests, read each "day" below as a week. For several jobs at once, see Several chores a day.

With 4 people and 30 days, one person averages 7.5 turns, give or take about 2.4, and lands between 4 and 12 in 94.10% of months. In a typical month the busiest person ends on 10 turns and the luckiest on 5.

The gap between the busiest and the luckiest is the number people notice. With 4 people it averages 5.61 turns and is 7 or more in 33.5% of months (simulated: 33.5%). In 62.4% of months everyone has had a turn and the busiest has still had at least twice as many as the luckiest.

From 3 to 5 people the average gap barely moves (5.3 to 5.7 turns), but with fewer turns each it weighs more.

Longer stretches don't even it out

Random does "even out over time," but in proportion, not in chores.

The gap grows roughly with the square root of the number of days. For 4 people the busiest person is expected to be 5.6 turns ahead after a month, 9.8 after three and 19.7 after a year, and a year ends 20 or more apart 47.2% of the time (simulated over 20,000 years: 46.5%). As a share of each person's total the gap shrinks, but a kid keeping score counts chores, not shares.

Streaks and long waits

Repeats are the other complaint. Someone gets the chore 3 days running in 99.75% of months with 2 people, 92.85% with 3, 77.9% with 4 and 62.6% with 5. With 4 people a run of 4 days or more happens in 28.4% of months. Each spin is still fair; it is 30 spins together that make some run likely. The streaks study shows why these feel rigged when they are not.

The flip side is a long break. Count the days between two turns of the same person: with 4 people over 30 days, the longest break in the household averages 11.3 days, and someone goes 10 days or more without the chore, while the others cover it, in 67.7% of months (27.7% with 3 people, 88.8% with 5).

Reweighting helps a little

The chore wheel page suggests, for a weekly job such as the bathroom, giving whoever did it last time *1 and everyone else *2, edited by hand. We tested re-weighting that way before every one of 30 daily spins with 4 people:

• The same person twice running drops from 1/4 to 1/7 of spins.
• 3 days running drops from 77.9% to 40.2% of months.
• A gap of 7 or more drops from 33.5% to 21.3%, and the average gap from 5.61 to 4.87.
• But at least one back-to-back still happens in 98.86% of months.

It softens the extremes without making the month even, and it is more work than the next fix.

Remove the winner: rounds that even out

Turn on Remove the winner after each spin and the wheel deals turns in rounds, a draw without replacement. Each spin takes the winner off; the wheel will not spin with one name left, so that person takes the round's last turn, and Reset puts everyone back.

Each round is a random order of everyone, and that one change fixes the count:

• Gap. After every complete round everyone has the same count, so the gap is 0 when the days divide evenly into rounds and exactly 1 otherwise. With 4 people over 30 days that means 7 or 8 turns each, a gap of 1.
• Streaks. Never more than 2 in a row: each person appears once per round.
• Waits. With 4 people, a person's next turn comes after 0 to 6 days off, 3 on average and 3 most often (25.0%). With daily spins and the winner left on, someone goes more than 6 days between two turns in 97.94% of months.

The back-to-back that rounds still allow

One repeat survives. The person who ends one round starts the next with chance exactly one over the number of people: 1/2 with 2 people, 1/3 with 3 people, 1/4 with 4 people, 1/5 with 5 people. With 4 people a 30-day month crosses 7 round boundaries, so at least one back-to-back happens in 86.7% of months (simulated: 86.7%), 1.75 on average. It is never three.

If even that is too much, press Shuffle once and go down the list in that order, over and over. With 4 people everyone then waits exactly 3 days between turns, with no back-to-backs. The randomness is used once, on the starting order. The same trade-off between fresh draws and a fixed rotation comes up with random partners week after week.

The wheel keeps a round only while the page is open: reloading brings back the example list, and typing in the list ends the round. Drawing a whole round at once gives the same odds as one spin each morning with removal on, so draw it and post the order on the fridge. To spread a round over several days instead, press Save on the names still left and load them the next day; save the full list too, because Reset after loading brings back the example chores.

Several chores a day

Now hand out 3 chores (dishes, trash, laundry) to 3 kids every day for a month.

Spinning each chore separately, winner left on, means each chore goes to a random kid. On any day some kid ends up with two or more chores 7/9 of the time (77.8%), one kid gets all three 1/9 of the time, and a given kid gets none 8/27 of the time. Over 30 days, 90 chores with a fair share of 30, the busiest kid ends 7 or more ahead in 67.6% of months (simulated: 67.7%).

Dealing the chores fixes the totals. Either use rounds (removal on, two spins and the third kid takes the last chore, then Reset), or the random team generator with the kids as names, Number of teams set to 3 and the chores typed into Team names (optional). Every kid gets exactly one chore a day and 30 in the month.

Dealing does not fix who gets which chore: under all three methods the dishes go to a random kid each day, so one kid washes up at least 7 more times than another in 31.1% of months. Only a fixed rotation, shifting one place each day, gives each kid each chore exactly 10 times.

When chores don't divide evenly

With 5 chores for 3 kids, the method matters more:

The team generator deals round-robin, so the first teams get the extra names. With the chores as names and the kids typed as team names in the same order every day, the first two kids get 2 chores and the third gets 1, every day: 60, 60, 30 in a month. Rotate the order of the team names each day and it becomes 50 each.

Rounds carried from one day to the next also give exactly 50 each, but on the 10 days a month that span three rounds, one kid gets 3 of the 5 chores with chance 1/3, and it needs the page left open from day to day.

Weighting an older kid

Sometimes an older kid should do more. Turn on Weighted choices (how weights work) and write Older *2, and that line gets twice the slice of each sibling's.

• Two kids. The older one carries 2/3 of the turns on average: 20 of 30. In a given month the count runs from 16 to 24 (5th to 95th percentile), and the older kid ends with fewer turns than the younger 1.88% of the time.
• Three kids. The older one carries 1/2 (15 of 30), 11 to 19 in a given month, and ends behind at least one younger sibling 7.20% of the time.

The weight gets the average right and the month only roughly: the older kid's share misses its target by 10 or more percentage points in 33.3% of months with two kids and 36.2% with three.

With Remove the winner after each spin on, a weighted line leaves the wheel whole, so Older *2 still has exactly one turn per round. The weight only moves that turn earlier: with three kids the older one goes first 1/2 of the time, second 1/3 and last 1/6. If you want exactly twice the turns, leave weights off and write the older kid's name on two lines with removal on. Each line is removed on its own, so every round gives the older kid two turns and each sibling one. Those two turns can land on back-to-back days, and across two rounds even three or four in a row. If that matters, post a fixed order such as Older, B, Older, C instead.

Setting up a fair chore wheel

The rule of thumb:

• Spin freely for one-off jobs where nobody keeps score: who clears up after the party, who picks the movie. Unevenness builds up only when the same job repeats.
• Use rounds for anything that repeats: everyone gets a turn before anyone gets a second.
• Use a fixed rotation when it matters who gets which job, or when back-to-backs would cause an argument: shuffle once, then rotate.
• Use weights for a deliberate imbalance, knowing that a month will only roughly match it, or put a name on two lines with removal on for an exact one (back-to-back days included).

Rounds on the chore wheel, step by step

1. Open the chore wheel and replace the example chores in What's on the wheel with your names, one per line. Save keeps the list in this browser for next time.
2. Under Wheel options, turn on Remove the winner after each spin.
3. Spin until one name is left; that person takes the last day. The wheel then shows "Add at least two choices to spin", which is expected. The Results panel lists every spin, and Copy exports them; add the name left on the wheel at the end.
4. Post the order, one name per day. For the next round, press Reset to put everyone back. If the list was edited or the page reloaded, Reset brings back the example chores instead, so load your saved wheel and check that removal is still on, since a reload can switch it off.

For the quickest version, press Shuffle after step 1 and use the list order as the rotation.

In a classroom, the random student picker already removes each student once picked. If winners stay on the wheel instead, how many spins it takes until everyone gets a turn shows how long the last names wait.

How to check these numbers

The script and its raw results are linked at the top of the page. Save the script and random.js in one folder and run node how-uneven-does-a-chore-wheel-get.mjs random.js results.json with Node.js 18 or later (about 170 seconds on our PC with Node 24; older versions are slower). The exact values will match this page; the simulated ones will differ in the last decimals.

• Exact values. One person's count is binomial. The gap comes from counting every way a month's turns can be split (a multinomial count in whole numbers), runs and waits from step-by-step chains over the days, and every round order is enumerated. Gaps beyond 30 days use the same exact methods in double precision.
• Simulation. The script loads the real random.js. A wheel spin is one CueRandom.pickWeighted() call over the lines still on the wheel, and a team generator press is one CueRandom.shuffle dealt round-robin. The 102.3 million wheel spins and the 5.4 million shuffle draws of 2.1 million team generator presses make the 107.7 million under the title; the 12.1 million turns handed to a round's last name without a spin are not counted.
• Checks. A simulated value passes when it is within four standard errors of the exact one: all 107 agreed, and the largest deviation was 2.73 standard errors. All 13 fixed results, such as a rounds gap of at most 1, held in every simulated month.

The random code itself is tested at scale in the fairness report.