Last updated . Source: SnagRes availability record, 61,530,416 checks across 284 restaurants through 19 October 2026. How we measure. Open dataset.
Neither, on average. Across the 284 restaurants where we measure both, a party of two finds a bookable table on 49% of dates and a party of four on 50%. The common advice to always book the smallest party is not supported by the measurements. What matters is the individual restaurant: at 24 of them four is materially easier, at 18 it is materially harder, and at 100 the difference is under five points.
Cote is the clearest: 70% of dates for a party of two, 100% for a party of four. Minetta Tavern runs 74% against 77%, Le Bernardin 61% against 77%, Chez Fifi 43% against 60%, and in San Francisco San Ho Won 47% against 93% and Quince 40% against 43%. Ten restaurants show 0% for two while still having tables for four, including Torrisi, Semma, Monkey Bar and Z & Y, which has never shown a two-top in 170,513 checks and shows a four-top on 77% of dates.
Dining rooms are largely built out of four-tops, and seating a couple at one costs the restaurant two covers. Some rooms therefore hold four-tops back for parties of four, which makes the larger booking the easier one. At counter restaurants and small rooms the effect reverses, because there a four-top is genuinely the scarcer object: Penny drops from 43% for two to 10% for four.
No, and nothing in our data supports it. Not one of the 284 restaurants we measure showed availability for a party of two while showing none at all for four, so the scenario this move is meant to solve did not appear once. At the 24 restaurants where four is the easier search, splitting actively hurts you: at Cote it takes you from every date in the window to 0% of them.
We do not measure a party of one, so we will not put a number on it, and neither should anyone who has not measured it. What our numbers do show is that the premise behind the claim is unreliable, since going from two to four does not make booking harder on average. The advantage a solo diner really has is the hour: prime time is an average of 26% of what we see open, so the 4:30 PM and 10:30 PM slots that a group cannot coordinate are where a single seat gets taken.
It does not change the release date, which is set by the restaurant and runs from 7 to 31 days ahead across the 136 rooms where we have observed it. It does change which sizes survive the first minutes after a drop. At a room built on four-tops the two-top goes first, and at a counter the four-top goes first, so the useful move is to know which shape the room is before the release rather than during it.
Every reservation guide, including two of ours until we checked, tells you to book the smallest party you can. Two-tops are more numerous, the reasoning goes, so a couple will always find a table where four people cannot. It gets repeated in every roundup and almost nobody has measured it.
We measure availability separately for a party of two and a party of four at all 284 restaurants we track, across Resy, OpenTable, Tock and SevenRooms, and the comparison now rests on 61,530,416 availability checks. Across the 284 restaurants with both figures on file, the average share of dates with a bookable table is 49% for two and 50% for four. Four is not harder. It is, very slightly, easier.
That average hides the interesting part. At 100 of those 284 restaurants the two numbers sit within five points of each other, so party size does almost nothing at all. At 24 a four-top is materially easier. At 18 it is materially harder. The lever exists. It is just not the one-directional lever people describe, and using it correctly means knowing which restaurant you are dealing with rather than applying a rule.
This is also the clearest illustration of why we watch 284 restaurants instead of advertising tens of thousands. Table Alert says it covers 62,198 restaurants. Nobody covering 62,198 rooms can tell you that a four-top at Cote behaves nothing like a two-top at Cote, because that answer only exists if you have asked both questions at the same restaurant, every fifteen minutes, for months.
The clearest cases are large enough to change a plan.
Cote, the Michelin-starred Korean steakhouse in Flatiron, showed a bookable table on 0% of dates for a party of two and 100% for a party of four. Every single date in the window. Minetta Tavern runs 42% for two against 77% for four. In San Francisco, San Ho Won goes from 47% to 93% and Quince more than doubles, 40% to 43%. Le Bernardin, three Michelin stars and a room nobody describes as bookable, sits at 61% for two and 74% for four.
The reason is structural rather than mysterious. Dining rooms are built out of four-tops, and a restaurant that seats a couple at a four-top has burned two covers. Some rooms protect against that by holding four-tops for parties of four, which is exactly the behaviour that makes the larger booking the easier one.
None of these are marginal rooms where nobody is competing. They are the restaurants people write about giving up on, and the larger search is the one that was quietly working the whole time.
Cote: 70% for two, 93% for four.
San Ho Won (SF): 63% for two, 77% for four.
Minetta Tavern: 74% for two, 77% for four.
Le Bernardin: 61% for two, 74% for four.
Chez Fifi: 43% for two, 57% for four.
Quince (SF): 17% for two, 40% for four.
Musaafer: 20% for two, 43% for four.
Claud: 33% for two, 40% for four.
This is the finding worth acting on. At ten of the restaurants we track, the two-tops were all taken while a table for four was still there to book.
Z & Y in San Francisco is the extreme case. Ask for two and its advance book is shut, across 170,513 checks. Ask for four and it showed a table on 77% of dates. Same restaurant, same calendar, two completely different answers depending on the number in the party-size box.
Several of the others are names people abandon entirely. Torrisi and Semma both sit at 0% for two and 0% for four. Monkey Bar goes from 50% to 47%, Bistrot Ha from 13% to 13%, House of Prime Rib in San Francisco and Oriana in New York from 0% to 7%. Even I Sodi, Jeju Noodle Bar and The Four Horsemen, flat zero for a couple, turned up a four-top on 0% of dates.
If you have searched one of these for two people, found nothing for weeks and concluded the room is closed to you, your search was answering a narrower question than you thought it was. The four-top calendar is a different calendar.
Z & Y (SF): 0% for two, 77% for four.
Monkey Bar: 0% for two, 47% for four.
Bistrot Ha: 0% for two, 13% for four.
Torrisi and Semma: 0% for two, 0% for four.
House of Prime Rib (SF) and Oriana: 0% for two, 3% for four.
I Sodi, Jeju Noodle Bar and The Four Horsemen: 0% for two, 0% for four.
The lever runs the other way often enough that you should check rather than assume in either direction.
Penny drops from 43% for two to 10% for four, the largest gap we see in that direction. The Progress in San Francisco falls from 83% to 53%. Bong goes from 60% to 33%, Lilia from 63% to 47%, Forsythia from 70% to 53%, Nubeluz from 80% to 70%. These tend to be rooms with counters, small tables or a fixed seating shape, where a four-top is genuinely the scarcer object.
The practical reading is simple enough. At a counter restaurant or a small room, expect four to be harder. At a full dining room built on four-tops, expect the opposite. Neither is a rule you can apply blind, which is why we publish the numbers per restaurant rather than as a slogan.
Penny: 43% for two, 10% for four.
The Progress (SF): 83% for two, 70% for four.
Bong: 60% for two, 50% for four.
Lilia: 67% for two, 33% for four.
Nubeluz: 87% for two, 83% for four.
Four Kings (SF): 13% for two, 10% for four.
Not one of the 284 restaurants showed availability for a party of two while showing none at all for four. The situation people plan around, where the only way in is to book two separate two-tops, did not occur once.
Sixteen restaurants sat at 0% for both sizes, and those are a different problem entirely. No party size gets in through the calendar there, so the release itself or a cancellation is the only route, and the size of your group is not what is standing in your way.
So if you are four people, book as four. Splitting throws away the four-top advantage at the 24 restaurants where it exists, and at a room like Cote it turns a 100% chance of finding a date into a 0% one. Past four the calendar stops being the right tool at all, because most hard restaurants never place a six-top online, which is what our guide to large-party bookings deals with.
Run both searches. It takes ten seconds and it is the cheapest item on any reservation checklist. A two-person search that comes back empty is not evidence the restaurant is full.
Then flex the hour rather than the size. Across the 132 restaurants where we have enough slot-level data, prime time between 7:00 and 8:30 PM accounts for an average of 26% of everything we see open, and at 39 of them it is 10% or less. Party size moves your odds by about a point on average. The hour moves them far more, which is the argument our guide to booking for one goes through in detail.
One thing we would not advise: booking a four-top for two people to game the calendar. It works, and it takes a table from a party who needed it, and enough of that behaviour is a large part of why restaurants hold four-tops back at all. If you are two, book for two and set an alert instead.
That restraint is also the line between us and the services that sell the table itself. TableSwap charges $75 per successful reservation, the practice New York's 2024 reservation law was written to stop. SnagRes watches the calendar at the party size you asked for and texts you when a date opens. It never books, never holds and never resells a table, so nothing we do makes the room harder for the person behind you.
Search two and four before concluding a restaurant is unavailable.
If your group is four, book as four rather than splitting into pairs.
At counter and small-room restaurants, expect four to be the harder booking.
Move the hour before you move the party size. Prime time is a minority of what opens.
Do not book a larger party than you are. It works, it is rude, and it is why the gap exists.
Keep reading: Booking a table for one · Booking for six or more · Booking New York from out of town · The easiest night to book · Where to take a client to dinner · Which tables are hardest right now
Browse all restaurant reservation guides
SnagRes checks 284 hard-to-book restaurants across four cities and alerts you when one opens. Watching one is free and no card is needed: watch a table free in New York City, or last-minute tables in NYC tonight.