CS2 Case Drop Rates by Rarity: Odds From Mil-Spec to Knife or Gloves

Updated · 6 min read

CS2 case odds: 79.92% Mil-Spec down to 0.26% knife or gloves. See per-opening chances, cumulative odds, StatTrak™ rates and expected value.

Opened CS2 weapon case with tapering rarity-coloured light beams rising to a floating gold knife

Every standard CS2 weapon case uses the same rarity table: 79.92% Mil-Spec, 15.98% Restricted, 3.20% Classified, 0.64% Covert and 0.26% Rare Special Item (knife or gloves). The gold tier works out to about 1 in 385 per opening. Each opening is an independent roll, so that figure does not guarantee a result after any fixed number of cases.

What are the official CS2 case drop rates by rarity?

The published rates run from 79.92% for Mil-Spec down to 0.26% for the gold tier, and they apply to every standard weapon case. What changes from case to case is the item pool and its prices, not the rarity odds.

Descending glass steps in rarity colours showing how drop chances shrink from Mil-Spec to gold
RarityColourChance per openingAbout 1 inStatTrak™ version (10% of drops)
Mil-SpecBlue79.92%1.257.99%
RestrictedPurple15.98%6.31.60%
ClassifiedPink3.20%310.32%
CovertRed0.64%1560.064%
Rare Special ItemGold0.26%3850.026% (knives only)

Where do these numbers come from?

The rates became public in 2017, when China required loot-box disclosure and Valve published them through its Chinese publishing partner, Perfect World. Liquipedia lists Perfect World as Valve’s partner for publishing Counter-Strike 2 and Dota 2 in China. The underlying disclosure gives three decimals: 79.923%, 15.985%, 3.197%, 0.639% and 0.256%. The two-decimal figures are the usual shorthand.

Community sources report that the September 2023 move from CS:GO to CS2 did not change the rates. Valve has not announced a change, but it has not published an audit either. The wiki’s CS2 Rarity Tiers and Case Odds guide covers each tier in more detail.

Why do the tiers step down the way they do?

Each tier is about five times rarer than the next one down, while gold is only 2.5 times rarer than Covert (a 2:5 ratio). The arithmetic checks out: 79.92 ÷ 5 ≈ 15.98, 15.98 ÷ 5 ≈ 3.2, 3.2 ÷ 5 = 0.64, and 0.64 ÷ 2.5 ≈ 0.26.

What do the odds mean in a single opening?

A single opening is one independent roll, so earlier results never change the next one. There is no pity timer or bad-luck protection. Over 100 openings the averages are roughly 80 Mil-Spec, 16 Restricted, 3 Classified, 0.6 Covert and 0.26 gold, but real runs scatter widely around those averages.

Grid of blue glass cubes with one gold cube beside a closed case, illustrating a 1 in 385 roll

How is a tier split between skins?

The tier chance is shared equally among the items in that tier. If a case has two Covert weapon skins, the 0.64% is divided between them before StatTrak™, wear and paint seed are considered.

The gold tier is diluted the same way. A hypothetical pool of 50 knife finishes gives each one 0.26% ÷ 50 = 0.0052%, about 1 in 19,000. Pool size varies, and so does the item type. Most cases carry knives in the gold slot, while a minority carry gloves instead. The Dead Hand Terminal, for example, has gloves as its rare special items. Check the gold pool before you open anything, and see the knife types guide for the models.

Wear and pattern are separate rolls on top of the skin. They can move a price a lot, as the float values guide explains.

How many openings does it take to get a knife or gloves?

No number guarantees one. The chance of at least one gold item after n openings is 1 − (1 − 0.0026)^n, which gives roughly 50% at about 270 openings, 63% at 385 and 95% at about 1,150.

OpeningsAt least one Covert or better (0.90%)At least one gold (0.26%)Key cost at $2.49
108.6%2.6%$24.90
5036.4%12.2%$124.50
10059.5%22.9%$249.00
38596.9%63.3%$958.65
1,150about 100%95.0%$2,863.50

Calculated from the rounded rates above. Key costs exclude the cases themselves.

The 385 row shows why “1 in 385” misleads. After 385 openings, about 37% of players would have no gold item, about 37% exactly one, and about 26% two or more. Nothing is ever “due”.

Does a long dry spell mean the odds are wrong?

No. With a 1-in-385 top tier, long dry spells are normal. One Steam forum poster reported around 600 openings with 451 Mil-Spec, 95 Restricted, 25 Classified, 5 Covert and no gold. At 0.26% per opening, zero gold in 600 happens about 21% of the time, so this result does not suggest the odds are wrong.

Testing the gold tier properly takes far more data than one account or one clip. At 0.26%, even 4,500 pooled openings would hold only about 12 gold items on average. Screenshots and streaks are not evidence either way.

How does StatTrak™ change the odds?

StatTrak™ is a separate 10% roll on eligible items, so it never changes the rarity odds. When an eligible item can be StatTrak™, the StatTrak™ version has a 1-in-10 chance. A StatTrak™ Covert is therefore 0.64% × 10% = 0.064% per opening, and a StatTrak™ knife is 0.026%, about 1 in 3,850.

Gloves use the gold slot but are not StatTrak™. Souvenir packages are a separate system. They have no knife or gloves slot and their items cannot be StatTrak™. The StatTrak™ vs Souvenir guide covers the differences.

What does an opening cost, and what does it return on average?

An official opening costs the case plus a key, and the average return is below that cost. The key costs $2.49 and the container is bought separately. Case prices move with the market.

Since October 2019, keys bought in-game cannot leave the purchasing account, so they cannot be sold on the Steam Community Market or traded. A key is a sunk cost (see Valve’s Key Change post).

Expected value (EV) is the probability-weighted average return. This hypothetical case costs $3.00 in total, and the skin values are assumed:

TierChanceAssumed valueContribution to EV
Mil-Spec79.92%$0.10$0.08
Restricted15.98%$0.80$0.13
Classified3.20%$5$0.16
Covert0.64%$30$0.19
Rare Special Item0.26%$300$0.78
Totalabout $1.34

That is roughly 45% of the cost, and about 58% of the EV comes from a 1-in-385 outcome that most players never see. A 2026 ranking of real cases found none returning more than it costs, with the best around seventy percent. Other 2026 tallies differ by method and price date, but they agree that the worst cases return closer to 18% of their cost before fees. See GameTyrant’s case ROI ranking for one example.

Selling on Steam lowers returns further. Steam documents a 5% transaction fee and a 10% game-specific fee for Counter-Strike 2, which people usually sum to 15%. SteamVaults’ breakdown explains why the exact arithmetic differs slightly.

Key takeaway: The rarity table is the same for every standard case, but the expected return is not. It depends on the skins in the pool and their current prices, and independent analyses put it below cost.

How do the odds differ on third-party case sites?

Valve’s table describes Valve’s cases only. A third-party site sets its own odds, prices and item pools for each case, and its margin is the house edge: the skins in a case are worth less on average than the price of opening it. Expect the same long-run pattern as with official cases, a loss on average, even when the site is honest.

Read each case’s listed odds and item values before opening it, and do not assume two similarly priced cases are equivalent. You can compare pools and prices on the all cases page. The free case shows how an opening plays out without spending. For the official mechanics, see How CS2 Cases Work.

What does provably fair verify?

Provably fair lets you check that a result was not altered after you committed to it. The common design works like this:

  1. The site shows a hash of its secret server seed before you play.
  2. You set or confirm a client seed.
  3. A nonce counts each opening so every roll is unique.
  4. The result is derived from these inputs with a published algorithm, often HMAC-SHA256.
  5. After the server seed is revealed, you can recompute the hash and the roll yourself.

This proves the roll was not tampered with. It does not make the odds favourable, and the house edge still applies.

Note: 18+ only. Case opening with real money is entertainment, not income, and nothing here predicts or promises a win. Set a budget and time limit first, never chase losses, and contact a local support service if play stops being fun.

FAQ

What is the chance of a knife or gloves from a CS2 case?

It is 0.26% per opening, about 1 in 385. Each opening is independent, so the chance does not build up. After 385 openings the chance of at least one is about 63%, not 100%.

Do all CS2 cases have the same odds?

Standard weapon cases share the same rarity odds. The item pool differs, so the value of each tier differs too. Some cases have knives in the gold slot and others have gloves. Souvenir packages, sticker capsules and Armory collections use different systems.

What is the chance of a StatTrak™ knife?

About 0.026%, or roughly 1 in 3,850. It is the 0.26% gold chance multiplied by the separate 10% StatTrak™ roll. Gloves cannot be StatTrak™.

Is opening a CS2 case positive expected value?

Not at typical market prices. A 2026 ranking found no case returning more than its full cost, with the best around 70%. Third-party sites add a house edge on top, so the same pattern applies.

Does provably fair mean I will win?

No. It lets you verify that a result was generated honestly and not changed afterwards. It says nothing about how often you will win, and the house edge still applies.

How many cases do I need to open to get a knife?

There is no guaranteed number. About 270 openings give roughly a 50% chance, and about 1,150 give roughly 95%. You could still go well past that without one.

Put it to use on CSDROP

Open cases, battle other players and upgrade your skins, with provably fair results.