The maps Counter-Strike community servers actually run

Dust2 alone took 24.8% of all player-hours on the Counter-Strike 1.6 servers JoinGG measured in the seven days to 28 September 2026, and 26.0% on Counter-Strike: Source. CS2 servers belong to Mirage instead: de_mirage took 36.5%, with Dust2 second on 16.6%. Behind those few maps sits a very long tail. On CS 1.6 alone, 11,917 different map names turned up, and most of them held almost nobody.

Where the numbers come from

JoinGG's poller asks every server it lists for its status every 15 minutes, using the same A2S_INFO query that Valve's server browser sends. The answer includes the map that is loaded and the number of players. An hourly sample is stored for each server, and everything below comes from the 168 hourly readings between 10:00 UTC on 21 September and 10:00 UTC on 28 September 2026. A reading counts only if the server answered and reported a map.

A map's share of player-hours is the sum of the player counts on every reading that showed that map, divided by the same sum for the whole game. Put plainly, it is the share of all the time spent on these servers that was spent on that map. Bots count. A2S reports bots inside the player total, so a server that fills its empty slots with bots adds to its map's share whether or not anyone human is there. The figures describe the 1,587 CS 1.6, 602 CS:S and 861 CS2 listings on JoinGG, not every server that exists, and the methodology page sets out the rules in full.

The top five in each game

All shares are of player-hours, rounded. On CS 1.6, de_aztec tied de_dust at 1.3%, and on CS2 de_inferno tied the Mirage duel map at 2.1%.

On the two older games, Dust2 fills the first row of the table and a variant of it fills the second. Dust2 appeared on 767 of the 1,587 CS 1.6 servers at some point in the week, nearly half of them, and on 206 of the 602 CS:S servers. After that the CS 1.6 list thins out fast: de_inferno has 3.0%, and the other familiar names (de_dust, de_aztec, de_nuke, de_train) hold about 1% each.

CS:S has kept more of its hostage maps than 1.6 has. Three of its top five are hostage maps (cs_office, cs_italy and cs_assault), and cs_ maps overall took 14.4% of CS:S player-hours against 5.7% on 1.6. A CS:S version of Mirage, de_mirage_csgo_v2, also reached 2.1% there.

CS2 is much more concentrated. Mirage ran on 379 servers and Dust2 on 332, and the top five maps took 62.3% of player-hours between them. The 267 legacy CS:GO servers still listed look much the same, with Mirage on 29.9%, Dust2 on 23.2% and 67.4% for the top five. Below the top two, the CS2 list is mostly other defuse maps (de_anubis, de_inferno, de_nuke, de_ancient, de_cache, de_overpass) mixed with maps built for one purpose, such as awp_lego_2, duels_mirage_1x1 and aim_redline.

Modes, read from the prefix

A map's prefix is a rough label for the mode a server was running. Defuse maps (de_) took 57.6% of player-hours on CS 1.6, 56.8% on CS:S and 70.7% on CS2, so the rest is where the communities differ.

Zombie maps are the biggest difference. On CS 1.6, zm_ maps took 13.3% of player-hours, spread over 1,610 different zm_ map names, which makes zombie servers the largest block after plain defuse. Zombie escape (ze_) had 1.5% there. CS:S sits in between, with zm_ at 5.4% and ze_ at 3.6%. On CS2 the order flips: ze_ took 6.6% and zm_ only 1.0%, so on the newest game the zombie mode people play is mostly escape. If you are choosing between the two, the guide to CS 1.6's zombie modes sets out how they differ, and zombie escape has a history across three games of its own.

The smaller slices are particular to each game. CS 1.6 still spends 4.0% of its player-hours on fy_ maps, 2.7% on awp_ maps and 2.5% on gungame. CS2 has its own set: awp_ at 4.9%, jailbreak (jb_) at 3.3%, and aim_ and duels_ at 2.7% each.

Surf, bhop and kz: many servers, few players

The modes one person can play alone look different from everything else in the data. On CS:S, surf_ maps were loaded in 6.0% of all server-hours but held 2.8% of player-hours, across 1,043 different surf map names. Bhop had the same shape: 4.2% of server-hours, 1.9% of player-hours and 1,187 names. The legacy CS:GO servers show the widest gap, with surf maps running for 16.9% of server-hours and holding 3.3% of player-hours over 625 names. On CS2 there were 535 surf names for 2.3% of player-hours, and 157 kz_ maps for 0.6%.

The libraries are large because these modes are about working through many maps, usually sorted by difficulty, and the populations are small because a surf run or a climb is mostly done alone. For a player, that changes what a quiet server means: three people on a surf server can be a normal evening, and the introduction to surf servers covers what to expect from one. If you want one particular map rather than a mode, searching by map name works better than scrolling a list sorted by players.

What a custom map costs

Each map outside the game's stock set has to reach every player before they can play it, and the long tail is made almost entirely of such maps. On CS 1.6 and CS:S, getting it there is the server's job. Without FastDL, which means sv_downloadurl pointing at a web server that holds the files, the client pulls the map and everything that comes with it from the game server itself, slowly, when it connects or when the map changes. Zombie servers are the heavy case, since the mod usually brings custom models and sounds on top of the map itself. A player facing a long download on a server they have never tried will often pick another one, so setting FastDL up properly matters more the deeper into the tail your rotation goes.

CS2 has no FastDL. Custom maps are Steam Workshop items, which the server loads with host_workshop_map or takes from a collection set with host_workshop_collection. That moves the hosting to Steam, and the player still has to fetch the map before the first round.

Picking the anchor map, then the identity

For a general server, the table shows where players already are: Dust2 on 1.6 and CS:S, Mirage on CS2. That map is the safe anchor, and it is also the most crowded choice available, with 767 CS 1.6 servers running Dust2 at some point during the week. Being on Dust2 gets a server considered. It gives nobody a reason to choose it over the next Dust2 server in the list.

A rotation that holds people through a map change usually puts the anchor beside maps the same players know, and the shares suggest the order: de_inferno, de_dust, de_aztec, de_nuke and de_train on 1.6; de_anubis, de_inferno, de_nuke and de_ancient on CS2. There is more on ordering a rotation so people stay for the next map in its own guide.

A server built around a niche gets chosen for its mode, and the prefix figures give a rough size for each niche. Zombie maps (zm_) hold 13.3% of CS 1.6 player-hours and fy_ maps 4.0%; on CS2, zombie escape holds 6.6% and jailbreak 3.3%. A slice that small can still fill one good server, and the old fun maps show how long a mode lasts once it has regulars. Before settling on either route, see how many servers already run your anchor: the Dust2 page for CS 1.6 and the Mirage page for CS2 list the servers on it.