If you checked Tuesday's 6-3 Blackhawks win over the Maple Leafs expecting a late empty-netter, there wasn't one. Chicago's last goal, by Tyler Bertuzzi, came at 12:10 of the third period with nearly eight minutes still to play, and NHL.com's recap mentions no empty-net goal.

The rules leave room for it. The NHL allows six players per team on the ice but doesn't require that one of them be a goalie. A 2-1 loss and a 10-0 loss both earn zero points, so a team that's behind risks very little by going all in.

How often does pulling the goalie actually work?

It doesn't work often, but it works often enough to be worth trying. From 2013-14 through 2019-20, teams down one goal that pulled their goalie tied or won the game about 15% of the time, according to Meghan Hall's analysis at Hockey-Graphs. Down two, it worked about 1% of the time, 25 times in seven seasons. Down three, it never worked.

The cost is real, and it's lopsided. Using 2015-16 data, Cliff Asness and Aaron Brown found that pulling the goalie raised the trailing team's chance of scoring by 1.18% per 10 seconds, but raised the opponent's chance by 2.54%. Their paper says it "nearly quadruples" the opponent's scoring chance while not even doubling the trailing team's. That trade only makes sense because giving up a second goal costs nothing extra in the standings.

You may see much lower success figures elsewhere. A Deseret News piece citing 32 Thoughts: The Podcast counted 34 games in 2023-24 where a team tied it in the last three minutes with an extra attacker and went on to win. That's about 2.6% of 1,312 games. But it's a share of all games, not of games where a goalie was pulled, so it makes pulls look less successful than they are. Each source defines success and picks its games differently, so their numbers can't be compared directly.

When should a team pull the goalie?

The research says earlier than most coaches do it. Asness and Brown recommend pulling with 5:40 left when down one, 11:40 when down two and 17:40 when down three. Alex Galea's model of 2003-2019 data puts the best time at a less extreme 3:00 left, with roughly 1-in-4 odds of scoring. In his data, pulls with 1:30 to 5:00 left worked 16±3% of the time, compared with 10±1% for pulls with under 1:30 left.

The recommendations don't fully agree. Galea cites the Asness and Brown paper as saying 6:10, not 5:40. Either way, every model says the best time comes minutes before most coaches pull.

Why do coaches wait so long?

Mostly because a coach who pulls early and loses takes the blame. Asness and Brown argue that "losing ugly can be career suicide", so coaches stick with the safe-looking choice of pulling late.

Habits have changed, just slowly. Galea found that average pull times crept from about 1.2 to 1.3 minutes left between 2003 and 2013, then reached 1.9 minutes by 2019, about 45% earlier. Hockey-Graphs found that the average pull for a one-goal deficit moved only about 30 seconds earlier over seven seasons, still well short of what the models recommend. Pulls that worked tended to start a little earlier.

Pulling itself is near-universal. Teams that trailed in the last two minutes pulled the goalie in 98% of those games. Still, teams spend only about 34 minutes per season with the net empty. There's no newer public league-wide average, so it isn't clear how close today's coaches get to the three-to-six-minute range.

Do the Maple Leafs and Blackhawks pull their goalies aggressively?

Toronto has a history of it. From 2013-14 to 2019-20, Toronto and Carolina were consistently more aggressive pullers than average. Chicago also pulled aggressively in 2015-16, a season when Patrick Roy's Avalanche got most of the credit for early pulls. Those figures are years old.

Are empty-net goals becoming more common?

Yes, though the rise seems to have leveled off. Regular-season empty-net goals went from about 200 a season in the early 2000s (about 0.17 per game) to 524 in 2024-25 and 508 in 2025-26, close to 0.4 per game, according to Jon Schwabish at PolicyViz. The rate barely moved between those two seasons, and it's too early to say where 2026-27 will land.

Sources don't always agree on the totals. The Deseret News counted 446 empty-net goals in 2023-24, and it's unclear whether the jump to more than 500 a season later reflects real change or a different way of counting.

Empty-net goals mostly go to good teams, because the teams leading late are the ones shooting at an open net. In 2025-26, Buffalo (27), Washington (25), Colorado (23) and Tampa Bay (23) scored the most.

Why do teams pull the goalie on a delayed penalty?

It's nearly free. When the referee signals a penalty, play continues until the penalized team touches the puck, and then play stops right away. The team that drew the penalty can add an extra attacker with almost no risk of giving up a goal. Wikipedia credits Milt Schmidt's Bruins with the first delayed-penalty pull, on April 16, 1958, against Montreal.

The extra skater makes a big difference. Hockey-Graphs found that 6-on-5 play produced just under 6.5 goals per 60 minutes, and 6-on-4 nearly 12.

What happens if a team pulls its goalie in overtime?

It can lose the point it already earned. A team that pulls its goalie in overtime and gives up an empty-net goal takes a regulation loss and gives up the point for an overtime loss. That makes the move much riskier in overtime than in a game that's already lost in regulation.

Who first pulled the goalie in the NHL?

Boston Bruins coach Art Ross did it in 1931, pulling his goalie in a playoff game against Montreal with Boston down 1-0 late, according to Chicago Booth Review. It didn't work. Accounts differ on exactly how much time was left.

Be careful with Wikipedia here. Its Extra attacker page gets the goalie (Tiny Thompson) and the date (March 26, 1931) right, but it calls Ross the Maple Leafs' coach and the opponent the Black Hawks, in a tie game. Ross coached Boston, and Thompson was a Bruins goalie. The first goalie pull wasn't a Leafs-Blackhawks moment, even if Tuesday's score makes that version tempting.