The NBA All-Star format explained, and why it keeps changing

You cannot answer the question "how does the NBA All-Star Game work" without first asking which year. That is not true of the Super Bowl, or the World Series, or a Champions League final. It is barely true of any other fixture in professional sport.

Since 2018 the league has torn up and rewritten its midseason showcase five separate times. It has been a playground draft, a game with no clock in the fourth quarter, a straight conference match again, a four-team knockout to 40 points, and now a nation-against-the-world round robin. Each change was announced with a press release explaining why the last one was not working.

Here is the NBA All-Star format explained as it stands, followed by the full sequence of revisions and, more usefully, the reason the league cannot leave it alone.

The current format is three teams and four games

The 2026 game, played on Sunday 15 February at Intuit Dome in Inglewood and shown by NBC and Peacock, ran on a format the NBA and the players' association announced on 11 November 2025.

Two teams of United States players and one team of international players, the World team, compete in a round-robin tournament of four 12-minute games. Each team carries a minimum of eight players.

The sequence is fixed and worth reading slowly, because it is not a normal bracket. Team A plays Team B in Game 1. The winner of Game 1 plays Team C in Game 2. The loser of Game 1 then plays Team C in Game 3. After those three games every team has played twice. The two teams with the best records advance to the championship game, which is Game 4. If all three teams finish 1-1, the tiebreak is point differential across each team's two round-robin games.

• 3Teams in the round robin
• 4Games played on the night
• 12Minutes in each game
• 8Minimum players on each roster

The elegance of the design is that everybody plays exactly two games before the final, so nobody is knocked out after twelve minutes and nobody sits for an hour. The awkwardness is that the World team plays Games 2 and 3 back to back, while each United States team plays Game 1 and then one of the other two. Whichever team is designated C absorbs the tighter turnaround.

Work the tiebreak through and it turns out to be sharper than it sounds. Three teams, two games each, and the only ways the round robin can end are one team at 2-0 with the other two at 1-1 and 0-2, or all three at 1-1. In the first case the qualifiers are obvious. In the second, the league falls back on point differential across each team's two round-robin games, which means a team that loses its opening game by two and wins its second by fifteen is better placed than one that won by two and lost by fifteen. In a twelve-minute game, a fifteen-point swing is roughly one good stretch of transition basketball. The tiebreak therefore does something no All-Star rule has done before: it makes the margin matter in a game you have already lost.

The roster problem is handled with an explicit backstop. If the ordinary All-Star voting does not produce 16 United States players and eight international players, the commissioner selects additional All-Stars to top up whichever group is short. The release also allows the international group to include American players with ties to other countries if necessary. That clause is the league admitting, in writing, that the format depends on an outcome the voting does not guarantee.

How you become an All-Star, which has barely moved at all

Underneath five format changes, the selection machinery has been almost untouched. That is the part of the process nobody rewrites, and it is more intricate than the game it feeds.

Twenty-four players are chosen, 12 from each conference. Ten of them are starters, five per conference, and their selection is a weighted three-way vote: fans carry 50 per cent, current NBA players 25 per cent and a media panel 25 per cent. The other 14 are reserves, seven per conference, chosen by NBA head coaches.

The arithmetic behind that weighting is more careful than "add up the votes". After all ballots are counted, players are ranked within each conference and each position group inside each of the three voting groups separately. A player therefore has a fan rank, a player rank and a media rank. Those three weighted ranks are averaged to produce a score, and the best scores in each position group become starters. If two players in the same position group finish level, the fan vote breaks the tie.

Ranking rather than raw totals is the load-bearing choice. It means a player who wins the fan vote by an enormous margin gains nothing extra from the margin; he gains one first place. A landslide in the popular vote is worth exactly the same as a narrow win, which is how the league keeps a global fan base from simply overwhelming the other two constituencies.

Voting runs through the NBA App and NBA.com, one full ballot per member per day, with a full ballot comprising three frontcourt players and two guards from each conference. On designated triple-count days, votes cast in that 24-hour window count three times, which is a mechanism for concentrating attention rather than distributing influence.

50%25%25%

• Fans50%
• Current NBA players25%
• Media panel25%

The three voting groups are ranked separately and their weighted ranks averaged. Fan voting also serves as the tiebreak inside a position group.

Show the numbers

What decides an All-Star starter

Item	Value
Fans	50%
Current NBA players	25%
Media panel	25%

For 2026 the league made one change to the selection itself: the All-Stars were chosen without regard to position. The old ballot structure of guards and frontcourt players was a legacy of a game with five positions and two conferences, and it had been producing arguments for years about guards excluded by an arbitrary quota. Removing it costs nothing once the teams are not built by conference anyway.

• Starters, chosen by the vote41.7%
• Reserves, chosen by head coaches58.3%

Starters are decided by the weighted fan, player and media vote. Reserves are decided by the 30 head coaches.

Show the numbers

Who actually picks the 24

Item	Value
Starters, chosen by the vote	10
Reserves, chosen by head coaches	14

Coaches are assigned by record, against a published cut-off. Under the 2018 arrangement the job went to the head coaches whose teams held the best record in each conference two weeks before the game. Under the 2025 arrangement the cut-off was a named Sunday in early February, and because that format needed four benches rather than two, the head coach of each conference leader took an All-Star team while an assistant from each of those two staffs coached the other two sides.

One rule attached to that assignment does real work and gets almost no attention: a coach who worked the previous year's All-Star Game is ineligible for the next one. The reward passes down the standings to whoever is next. Without it, a dominant team's coach would collect the honour every February for as long as the run lasted, and the assignment would stop being a recognition of a season and start being a fixture. The ineligibility rule is the same idea as a rotating chair, applied to a job nobody would otherwise think to rotate.

Replacements are the commissioner's call. A withdrawn All-Star is replaced by appointment, not by the next name on a list, which is why replacement selections are argued about with a heat the original selections rarely generate.

The voting calendar is a broadcast schedule with a ballot attached

Read the published voting timetable and the shape of the whole event becomes clear, because every date on it is also a television date.

For the 2025 game, voting opened on Thursday 19 December at 10am Eastern and closed on Monday 20 January at 11:59pm. Inside that five-week window sat six days on which every ballot counted three times: 21 December, Christmas Day, and then 3, 10, 17 and 20 January. Fan voting updates were published on each of the first three Thursdays in January. Starters were revealed on the Thursday after voting closed, and reserves the Thursday after that.

Lay those out and there is a scheduled news event roughly every seven days for seven consecutive weeks. Christmas Day is the league's largest regular-season broadcast window and it is a triple-count day. The remaining triple days fall on Fridays and on the final day of voting, which is when a campaign either closes a gap or does not. The weekly returns exist so that the standings can move in public rather than being revealed once at the end.

None of this is hidden and none of it is dishonest. It is the answer to a genuine design problem: an open global fan vote with no structure would be effectively decided in its first 48 hours by whoever mobilised fastest, and everything after that would be dead air. Spacing the multipliers keeps the result live for five weeks and gives smaller markets a series of moments at which a concerted push can still change something.

The side effect is that being voted an All-Star is now partly a campaigning exercise. Clubs run them, broadcasters cover them, and a player with a large following in a populous country has an advantage that has nothing to do with basketball. The 50-25-25 weighting exists precisely to bound that advantage: however lopsided the fan vote, it can only ever supply half the score, and the player and media ballots are cast by people who have watched the season.

Five formats in nine years

The revisions are easiest to understand as a sequence, because each one is a reaction to the failure of the one before it.

2018: the captains draft. Announced in October 2017, this ended conference against conference for the first time in the game's history. The two captains were the starter from each conference who received the most fan votes, and they drafted their teams from the whole pool of 24 selections without regard to conference affiliation. Selection stayed conference-based; the teams did not. Each side also played for a charity, which was the first appearance of the money lever.

2020: the Final Target Score. Announced in January 2020, this is the change that actually worked as basketball. The first three quarters were played as normal 12-minute periods, except that each one started fresh at 0-0. At the start of the fourth quarter the game clock was switched off and a Final Target Score was set, calculated by taking the leading team's cumulative score across the three quarters and adding 24, the number Kobe Bryant wore for the last ten seasons of his career. The teams then played an untimed fourth quarter, and the first to reach the target won.

The league published its own worked example. If the cumulative score after three quarters is 100-95, the target is 124. The team on 100 needs 24 more points before the team on 95 scores 29, and the other way round. With no clock, the game can only end on a made basket or a made free throw.

The charity structure was welded onto the same frame. Each of the first three quarters was worth $100,000 to the winning team's chosen Chicago organisation, and winning the game itself was worth $200,000. A tied first or second quarter rolled its award into the next quarter; a tied third quarter rolled into the game award. A team that won all three quarters and then hit the target first delivered $500,000 to its charity and $100,000 to the other side's.

2024: the classic returns. Announced in October 2023, this restored the Eastern Conference against the Western Conference, four 12-minute quarters and ordinary cumulative scoring. The untimed fourth quarter and the target score were dropped. Captains were still named, defined as the starter in each conference with the most fan votes, and the teams still competed to win each quarter for a charity. It lasted one year.

2025: the mini-tournament. Announced in December 2024, this was the biggest structural break of the lot. Four teams of eight players, three games, and each game won by the first team to reach or pass 40 points. Two semi-finals, then a final. Three of the four teams were drafted live on television by Charles Barkley, Shaquille O'Neal and Kenny Smith as honorary general managers of Team Chuck, Team Shaq and Team Kenny. The fourth team was the winner of the Rising Stars event two nights earlier, with Candace Parker as its honorary general manager.

It also attached real money to the outcome for the first time. A prize pool of $1.8 million was split so that each player on the winning team received $125,000, each player on the runner-up $50,000, and each player on the third and fourth-placed teams $25,000.

Winners125k

Runners-up50k

Third25k

Fourth25k

Per player, in thousands of US dollars, from a $1.8m pool across four teams of eight.

Show the numbers

What a player earned per finishing position in the 2025 mini-tournament

Item	Prize per player
Winners	125k
Runners-up	50k
Third	25k
Fourth	25k

2026: United States against the World. Announced in November 2025, and described above. It abandons both the draft and the conference and replaces them with the one dividing line the league had never used.

The untimed fourth quarter was the best of the five, and it still went

The target score deserves a defence, because as a piece of rule design it solved a problem that ordinary basketball has never solved and the league dropped it anyway.

Think about what a clock does to the end of a close game. The trailing team fouls deliberately, because a foul stops the clock and converts an opponent's possession into two free throws instead of a shot at the rim. The leading team, meanwhile, is trying to use time rather than score with it. Both behaviours are rational, both are tedious, and between them they turn the most important three minutes of a basketball game into a sequence of free throws and timeouts.

Remove the clock and both behaviours vanish at once. Fouling is now purely self-harming: the free throws it concedes move the opponent closer to the target, and there is no clock left to stop. Running down time is impossible for the same reason, because time no longer exists as a resource. The only way to win is to score the required points, and the only way to stop the other team is to actually get a stop. A game played to a target cannot end on a whistle, a turnover or an expired clock. It ends on a made basket or a made free throw, every time.

The choice of 24 as the increment was a memorial rather than a mechanism, but the size of the number matters more than the sentiment. Too small and the final period is over in three possessions before anyone has settled. Too large and it becomes a full quarter with the clock hidden, which defeats the point. Somewhere around twenty to thirty points is the band where the finish is long enough to swing and short enough to be urgent.

What killed it was almost certainly not the basketball. It was the scheduling. A period with no clock has no known length: it might last four minutes or twenty, and neither the arena nor the broadcaster can plan around that. A live event that cannot tell its network when it will finish is a problem, and it is the one problem an exhibition has no argument against, because the whole justification for the event is the audience.

The mini-tournament and the round robin are, read in that light, attempts to keep the good part and lose the bad one. Twelve-minute games and races to 40 both compress the stakes into a short window while leaving the running time predictable. The target score got there first and got there better; it just could not tell anyone when it would be over.

The three levers, and the one the NBA had not pulled

Strip the announcements of their language and every change since 2018 pulls one of three levers.

The first is who your teammates are. The draft format existed to make the matchups novel: teammates against teammates, a superstar picking a rival, a bench built by hand. Novelty in the roster is cheap to produce and it generated a genuinely good television segment in the draft itself. What it did not do was change anybody's behaviour once the ball went up, because a team assembled forty minutes ago is not a side you would take a charge for.

The second is what winning is worth. Charity money arrived in 2018, per-quarter charity money in 2020, and direct prize money in 2025. This is the lever with the clearest theory behind it and the weakest evidence. Professional athletes with nine-figure contracts are not motivated by $125,000, and the league surely knows it. What prize money does is give the broadcast a legitimate reason to describe the game as a competition, which is a different objective and a defensible one.

The third is how short the unit of competition is. Twelve minutes is short enough that a bad three minutes cannot be shrugged off. A race to 40 is short enough that a run of two or three baskets decides it. The 2025 and 2026 formats both rest on the same insight: the problem with a 48-minute exhibition is not the players, it is the length. Nobody defends hard for 48 minutes in February for nothing, but almost anybody will defend hard for twelve minutes when the alternative is being eliminated in front of a full arena.

The fourth lever, the one the NBA left alone until 2026, is identity. This is the only mechanism that has ever reliably produced effort in an exhibition, and every other sport already knew it. Golf gets its most competitive week of the year out of a match-play event with no prize money because the players are wearing a continent's colours. Cricket, rugby and football all found long ago that a badge people care about does more for intensity than any incentive scheme.

The reason the NBA could not use it before is arithmetic. A World team needs eight international All-Stars, and for most of the league's history there were not eight. The format exists now because the composition of the league changed underneath it, which makes the 2026 game a milestone about the sport rather than about the event.

1. 1The weighted starter voteFans carry half, current players a quarter, the media panel a quarter. Each group is ranked separately and the weighted ranks are averaged. Ten starters, five from each conference.
2. 2The coaches' ballotThe 30 head coaches choose 14 reserves, seven from each conference, and for 2026 they did so without regard to position. That completes the 24.
3. 3Sorting into nationsThe 24 are split into 16 United States players and eight international players. If the vote does not produce that split, the commissioner appoints additional All-Stars until it does.
4. 4Building three rostersTwo United States teams and one World team, each with at least eight players. Where the commissioner has topped a group up, at least one team carries more than eight.
5. 5The round robinTeam A plays Team B. The Game 1 winner plays Team C, then the Game 1 loser plays Team C. Twelve minutes each, so every team has played twice.
6. 6The championship gameThe two teams with the best records meet in Game 4. If all three finish 1-1, point differential across the two round-robin games decides who plays.

The selection half of the pipeline is conference-based. The competition half is not. Bridging those two facts is what the commissioner's top-up power is for.

What the conference-based selection now costs

There is an unresolved seam running through the current arrangement, and it is worth naming because it will produce an argument sooner or later.

Players are still selected by conference. Twelve from the East, twelve from the West, with starters ranked inside conference position groups and reserves chosen by coaches voting within their own conference. The game is then played by nationality. Those two facts have nothing to do with each other.

The consequence is that the pool of American All-Stars and the pool of international All-Stars are both by-products of a process that was not trying to produce them. A conference could plausibly return a heavily international group of selections while the other returns almost none, and the split would still have to come out at 16 and eight. That is exactly why the commissioner's top-up power exists, and why the release specifies that an American player with ties to another country may be moved across if necessary.

Any appointment made under that power is, by construction, a player the voters and the coaches did not select. The first time a genuinely borderline case is added to make the numbers work, the discussion will be about whether he deserved it, and the honest answer will be that deserving was never the criterion.

The cleaner fix would be to select by nationality in the first place: vote for a United States squad and a World squad directly. The reason the league has not done that is that the conference structure is what makes the selection feel like an honour attached to a season, and the honour is the part players actually care about. Being named an All-Star is a line in a career record and a clause in a contract. Which shirt you wear for 48 minutes in February is not.

The things that never change

For all the churn, a short list of features has survived every rewrite, and the list tells you what the league considers untouchable.

The number 24 has not moved. Twelve per conference has not moved. The 50-25-25 weighting has not moved since it was introduced. Head coaches still choose the reserves, and they still cannot vote for their own players. The commissioner still appoints replacements. Coaching assignments still follow regular-season record.

And the game still counts for nothing. Every experiment has been an attempt to make an exhibition feel consequential without giving it a consequence, because a consequence would mean either home advantage in the finals, which the league tried nowhere, or a roster spot in a competition that matters, which nobody has proposed.

That constraint is not going away, and it is worth being clear about why. The All-Star Game sits in the middle of an 82-game season in which availability is the scarcest resource on any roster. A format that genuinely raised the injury risk would be opposed by every team in the league within a week. Every revision has had to be entertaining and safe at the same time, and those two requirements pull against each other. Twelve-minute games are, among other things, an answer to that: they raise intensity while lowering total minutes on the floor, which is the only way to have both. The same calculation shapes how teams manage minutes across a long season, and it is the reason a February exhibition is negotiated rather than imposed.

How the NBA's problem compares

The NBA is not alone here, and the comparisons are instructive because the other leagues have failed in the same directions.

American football's all-star fixture eventually gave up on being a football game at all and became a skills competition, which is a more radical answer to the same problem and arguably a more honest one. If the game cannot be played properly, stop pretending it is a game.

The other route is to invent a competition that matters and put it inside the season, which is what the league did with its midseason knockout event. That approach accepts that the All-Star Game will never generate real intensity and builds a separate thing that can. It is also the reason the All-Star weekend now has less work to do: it no longer has to be the only thing standing between Christmas and the scramble for the last playoff places.

Broadcast economics run underneath all of it. An All-Star Sunday is a large, scheduled, predictable audience, and rights deals are written around exactly that kind of inventory. Four short games with a knockout structure produce more natural breaks, more segments and more moments a highlight package can be built from than one 48-minute exhibition does. Nobody at the league says the format is designed for television, and nobody needs to; the way modern rights agreements are structured makes the incentive obvious enough.

Reading the next format change before it is announced

There will be another one. On the evidence of the last nine years, the base rate is roughly a change every other season, and the 2026 format solved a problem the previous four did not touch, which usually means it survives longer than they did.

When the next announcement arrives, four questions will tell you whether it is a real change or a rebrand.

Does it change who is on the floor in the last two minutes? Every format that mattered did. The target score did it by removing the clock, so nobody could run out the game. The mini-tournament did it by making elimination immediate. A change that leaves the closing minutes looking the same is decoration.

Does it shorten the unit of competition? Twelve minutes and a race to 40 both work for the same reason. A format that returns to 48 minutes of continuous play is betting on motivation rather than structure, and that bet has been lost repeatedly.

Does it give the players something to belong to? Charity money is a reason to try. A flag is a reason to care. The distance between those two is the distance between the 2020 format and the 2026 one.

Does it survive contact with the selection process? Any format that needs a particular composition of All-Stars, whether by nationality, position or anything else, is depending on an outcome the vote does not control. The more appointments the commissioner has to make to force the shape, the more the honour and the event come apart.

Watch those four and the annual announcement stops being a surprise. For everything else the league does with its calendar, the basketball archive is the place to start, and the way the modern game's shape was set by a single line on the floor is covered in the story of the three-point arc.