We tested 114 sweepstakes casinos. Here's who actually pays.

How the numbers narrow

Each step below is a smaller set than the one above it, and the pay rate is computed over the last two, not the first. Publishing only the top number would let a reader assume every operator we opened an account with was also payout-tested, which is not true and never was.

1. 114 operators tested first-hand, with a real deposit on a real account
2. 45 of those where we requested at least one cashout
3. 40 where that question has resolved: a payout we watched land, or a redemption that closed without paying
4. 32 that paid us at least once

Pay rate = operators paid at least once / operators with a resolved redemption outcome = 32 / 40 = 80%. 5 operators still have no resolved outcome and are held out of both sides of that fraction.

The numbers at a glance

114 operators tested first-hand · 45 we requested cashouts from · 40 with a resolved outcome · 5 still unresolved · 32 paid at least once · 220 redemption attempts logged · 105 delivered · September 25, 2025 – August 10, 2026

The top-line finding

Out of the 40 casinos we put real money into, asked to pay us back, and have since got an answer from, 32 delivered at least one cashout: 80%. That is both better and worse than the category's reputation deserves. Better, because most of the brands we tested do pay: the model works, and the legitimate operators honor it. Worse, because the gap between the ones that pay cleanly and the ones that stall is wide, and from the outside you usually can't tell which is which until your own money is the one waiting.

We logged $24,030.65 in redemptions where we recorded the amount, the single biggest being $5,772. The fastest clean cashout landed the same day; the slowest dragged on for days. The point of this study is to put receipts behind those words.

Who paid us most reliably

These are the operators that delivered the most confirmed redemptions to our accounts. "Delivered" means a payout we actually received, not a request we submitted and crossed our fingers over. The count is how many separate cashouts each one paid us during the study.

The standout for speed was LuckyHands: a $199 redemption that cleared the same day we asked for it: request to money in hand inside a single business cycle. The largest single payout we received across the whole study was $5,772. A high payout count is the cleanest possible signal in this category: it means we have personally watched the operator turn a balance into real money, more than once.

Requested, not yet confirmed

We're deliberately careful with this group. At these operators we have an open redemption request that we have not yet seen land. That is not a refusal and it is not a failure; many redemptions sit in a normal processing window before they arrive, and several of these brands score well on other evidence. We simply haven't watched the money hit our account yet, so we won't count it as paid until we do:

• Cashoomo: request open, payout not yet observed , though a separate redemption here already closed without paying, so it does count in the pay rate as an operator that has not paid us
• ChipNWin: request open, payout not yet observed , though a separate redemption here already closed without paying, so it does count in the pay rate as an operator that has not paid us
• Cider Casino: request open, payout not yet observed , and nothing else on the record here has resolved, so it is held out of the pay rate entirely
• HelloMillions: request open, payout not yet observed , and nothing else on the record here has resolved, so it is held out of the pay rate entirely
• JackpotGo: request open, payout not yet observed , and nothing else on the record here has resolved, so it is held out of the pay rate entirely
• Legendz: request open, payout not yet observed , though a separate redemption here already closed without paying, so it does count in the pay rate as an operator that has not paid us
• Spinfinite: request open, payout not yet observed , and nothing else on the record here has resolved, so it is held out of the pay rate entirely
• StormRush: request open, payout not yet observed , and nothing else on the record here has resolved, so it is held out of the pay rate entirely
• TaoFortune: request open, payout not yet observed , though a separate redemption here already closed without paying, so it does count in the pay rate as an operator that has not paid us

That split is the whole point of the denominator. An operator only enters the pay rate once one of its redemptions has actually resolved, either money we watched arrive or a request that closed without paying. Today 5 of the 45 operators we requested a cashout from have neither, so they are outside both sides of the fraction rather than sitting in it as operators that failed us. We'll update this section as those requests resolve, and the rate will move when they do.

Where redemptions failed or were a mixed bag

Some operators paid us eventually but made it harder than it should have been, and a few outright failed a redemption during testing. We're keeping this fair: several of the names below do pay some of the time. A mixed record isn't the same as a refusal to pay, but it's exactly the friction you want to know about before your own balance is on the line.

• RealPrize pays, but cancelled our redemptions repeatedly across December 2025 and January 2026, and quotes a turnaround of "up to 14 business days." It eventually came through, yet the repeated cancellations are the kind of pattern that turns a win into a chore.
• GoGoGold cancelled our redemptions multiple times, and the support channel was an unhelpful AI bot that couldn't explain why. Mixed at best.
• Dara failed a redemption during testing and carries a 60-day Sweeps Coin expiry policy, a clock on your balance that quietly works against you if a cashout stalls.
• LoneStar failed a redemption on what presented as a debit-card processing issue. We couldn't get it to complete.
• FortuneWheelz returned a flat "we are unable to process your request" on a redemption attempt, with no path forward offered.
• Legendz, Scarlet Sands, and SweepJungle each cancelled or returned a redemption rather than paying it out on the attempt we made.

None of this is a one-off bad review. We're reporting redemptions we personally submitted and watched stall, get cancelled, or get bounced. Where a brand also paid us on a different occasion, we've said so; the takeaway is to start small and confirm one clean cashout before you trust any of them with a real balance. Review complaints are the sibling evidence stream to failed redemptions: what other players report across the whole catalog, and which operators it has cost points, lives in our complaint index.

Documented: the Zula bank-statement rejection

The clearest single record we have of that wall comes from Zula Casino. During account verification we submitted a bank statement as proof of residency, and it was declined across two dated support tickets on consecutive days. We are presenting this as a documented record of one player's verification experience, with the operator's own wording quoted from the transcripts, not as a claim about Zula's intent.

On the first ticket (1887226, April 30, 2026), Zula declined to proceed, stating we regret to inform you that we are not able to continue with the account verification and we require an actual statement from a Bank. In the same exchange it stated its policy plainly: we don't accept bank statements as proof of residency. On a second ticket the next day (1890844, May 1, 2026), with further documents submitted, the rejections continued, this time on the grounds that the attached file does not match the details on your account.

We are quoting only Zula's own words and drawing no conclusion about why. The point for a reader is narrower and useful: a routine residency document was repeatedly declined, and a redemption you cannot get verified is a redemption you cannot collect.

One request, eleven brands: the UTech Solutions network

The "separate brands" illusion breaks down on the corporate side too. Using a single consolidated data-access request, we contacted eleven sweepstakes brands as properties of one company, UTech Solutions LLC: JackpotRabbit, SweepShark, ScarletSands, VegasWay, Playtana, MrGoodwin, Sweepico, FireSevens, DexyPlay, StormRush, and NoLimitCoins.

To be precise about the evidence: we filed one CCPA/CPA request naming all of them together as UTech Solutions LLC properties and delivered it to working privacy and support addresses at those brands. We are reporting who we contacted and how they responded, not a corporate filing. What the responses showed is hard to dismiss. Several brands returned the identical it is in your account, no export, no tax form reply, and the same support agent, who identified himself as James, answered for two of the brands within about a minute of each other on the same day.

A shared support team and a word-for-word identical script across nominally separate brands are documented, dated signals of a common backend, whether or not every brand shares one legal owner. For players the takeaway is the same as with the KYC wall: many apps that look independent are not, so a problem at one can quietly be a problem at all of them.

Verification and speed, measured

Across the study we ran 52 first-hand identity checks and cleared 47 of them, a 90% verification rate. The ones that didn't clear weren't random; the shared-backend cluster above accounts for the rejection pattern. KYC is the single biggest reason "delayed" payouts get delayed, which is why we measure it separately from payout speed.

On the 13 delivered redemptions where we have a clean request-to-received clock, the fastest landed in 0h (same-day), the median sat at about 24h, and the slowest took 96h. Speed tracks with payment method, request size, and whether verification was already cleared: finish KYC before you have a large balance waiting and you remove the most common cause of a slow cashout.

Methodology

This study is built entirely on first-hand testing, not aggregated reviews. Here is exactly how the numbers were produced:

• Real money, real accounts. Every test used a genuine deposit and a real identity (Noah's own accounts), not a press account or a sandbox.
• "Delivered" means received. A redemption only counts as paid when the money actually arrived (recorded as a received timestamp). A submitted request on its own counts for nothing.
• Speed is request-to-received. Payout speed is measured from the moment we submitted the redemption to the moment the funds landed, on the subset of delivered redemptions where we have both timestamps.
• Failures are documented, not counted as paid. Cancelled, returned, or refused redemptions are reported in full above, but they never inflate the pay rate or the dollar total. They do count as a resolved outcome, because a redemption that closed without paying is an answer.
• Pending stays pending. Open requests we haven't watched land are held out of the headline pay rate until they resolve, on both sides of the fraction. An operator whose redemptions are all still open is not counted as an operator that failed to pay us; today that is 5 of the 45 we requested a cashout from.
• The pay rate is operators, not redemptions. It is operators paid at least once divided by operators with a resolved redemption outcome, 32 / 40 on this build. One operator that paid us six times counts once, and so does one that cancelled on us twice.
• Bound to the database at build. Every count, percentage, and amount on this page is read from our first-hand-testing database when the page is built, so it can't drift from the underlying records. The dollar figures in the net-loss section above come from the operators' own data exports and our purchase-and-redemption receipts, which is why they are stated as sourced totals rather than database-bound counts.
• Affiliate-funded, not affiliate-weighted. BonusBandit is funded by affiliate links, and we disclose it. The trust ratings are computed mechanically and are not affiliate-weighted, so a site cannot buy a better score, and the loss reported here is our own money.

Data stamp: 2026-08-10T17:49:07.450Z · testing window September 25, 2025 – August 10, 2026