If you are searching for the most addictive gambling games, what you are really asking is which products cause the most harm. The research points the same way in most countries. Electronic gaming machines — slot machines, pokies and fruit machines, in venues and online — and online casino games are tied to the most harm. Sports betting comes next, and is growing fast. Lottery draws sit at the bottom.
But no game is addictive on its own. Risk comes from how a product is built, how easy it is to reach, and who is using it. This page explains what the evidence shows, how researchers score a product’s risk, and which design features keep turning up in the high-risk group.
The short answer, product by product
The table below sums up where the main gambling products sit. It draws on the World Health Organization gambling fact sheet (2 December 2024), the Lancet Public Health Commission on gambling (October 2024), and national survey work described further down this page.
| Product | What the evidence shows | Design features behind the risk |
|---|---|---|
| Electronic gaming machines (slots, pokies, fruit machines) | Consistently linked to the most harm across countries | Very fast play, continuous sessions, near misses, losses dressed up as wins |
| Online slots and online casino games | Highest rates of problems among people who use them | All of the above, plus always open, no travel, card and wallet payment |
| Casino table games, including machine terminals | Raised risk in British and Australian data | Fast rounds, easy to repeat, stake size can jump |
| Sports and race betting, especially in-play | Second highest group globally; growing quickly | Many markets per event, bets settled in minutes, feels like skill |
| Scratchcards and instant-win games | Raised risk once other gambling is accounted for | Instant result, cheap to repeat, sold almost everywhere |
| Bingo | No clear independent link in recent British data | Slower pace, social setting, natural breaks |
| Lottery draws | Lowest risk of the main products | Long wait for the result, one purchase, hard to repeat quickly |
These are group averages. They describe patterns across thousands of people. They do not predict what will happen to any one person. Plenty of people are harmed by products near the bottom of that table, and most people who play slots are not harmed at all.
What people mean by “the most addictive gambling games”
The phrase is a shortcut, and it hides two problems.
The first is that the surveys behind these rankings do not measure addiction. Most use the Problem Gambling Severity Index, or PGSI. It is a nine-question screening tool. It flags people who may be at risk, so they can be offered help or assessed properly. The Gambling Commission’s own guidance on using its statistics states plainly that its figures should not be used “as a measure of addiction to gambling”. A diagnosis of gambling disorder is made by a clinician, not by a survey. We explain that difference in our guide to gambling addiction.
The second problem is that risk is never only about the game. It is the product, the person and the setting together. The same slot machine carries a different risk for someone playing once a month on a night out and someone playing alone at three in the morning after a bad week. What product research can tell us is which designs shift the odds — which ones make it easier to keep going, and harder to stop.
How researchers score the risk of a gambling product
Two formal tools are used to rate gambling products, and both work the same way: break the game into its parts, score each part, add it up.
GAMGARD
GAMGARD was built in 2006 by Dr Richard Wood and Professor Mark Griffiths, working with an international group of researchers and clinicians. An evaluation for New Zealand’s Ministry of Health, published on 8 November 2021, lists the ten structural risk factors it scores:
- Event frequency — how many rounds you can play per minute
- Multigame opportunities — whether you can run several games at once
- Fixed or variable stakes — how far you can push the bet size up
- Prize-back percentage — how much of the money staked is returned as prizes
- Jackpot size
- Near-win opportunities
- Continuity of play — whether anything naturally stops you
- Accessibility
- Payment options
- Illusions of control
Newer versions also credit four protective features: pop-up reminders, spend limits, time limits and feedback tools. We cover those in our guide to responsible gambling.
ASTERIG
ASTERIG was developed between 2006 and 2010 by the Wissenschaftliches Forum Glücksspiel, a German gambling research forum, and published in The Journal of Gambling Business and Economics. It scores a product against weighted risk indicators to produce a single risk figure, so that different products can be compared on one scale. It has been used in European regulatory debates about which products need tighter rules.
How much weight to give these scores
Not too much, on their own. The New Zealand evaluation of GAMGARD concluded that it works as “a preliminary screen, not a precision tool”, noted that its scoring weightings are not public, and warned that it “should never be used as a standalone decision-making tool”. ASTERIG has been through a similar exercise: a paper titled Challenges in the Measurement of Gambling Product Risk: A Critical Review of the ASTERIG Assessment Tool was published in International Gambling Studies in 2022.
That is worth knowing if you ever see a game described as “certified low risk”. These tools compare designs. They do not certify a product as safe.
The design features that raise risk
Speed
Speed is the feature researchers return to most often. A lottery draw gives you one result a week. An online slot can give you a result every few seconds. That changes how many decisions you make, how many losses you absorb in an hour, and how much chance you have to stop and think.
A critical review by Harris and Griffiths, published in the Journal of Gambling Studies in 2018, looked at eleven studies on speed of play. It found that faster games were rated as more exciting by everyone, that people already experiencing gambling problems showed a stronger pull towards rapid play, and that they found it harder to stop once a fast session was under way. The review was careful, though: effects on time and money spent were mixed across studies, and the authors argued that limit-setting and feedback tools may do more good than simply slowing games down.
Continuity
Some products end. You buy a lottery ticket and then you wait. Others do not end until you decide to end them. A machine will keep taking bets for as long as you sit there. Researchers call this continuous play, and it is the reason the gap between a weekly draw and a slot machine is much wider than the gap between two different slot machines.
Near misses and losses that look like wins
Two reels on the jackpot symbol and the third just above it is a loss. It does not feel like one. In the same way, staking £1 and winning 20p back is a loss of 80p, but if the machine plays a win sound and flashes lights, it can register as a win.
Regulators now treat these as design defects rather than harmless flourishes. In its package of online game design changes announced on 2 February 2021, the Great Britain Gambling Commission banned sounds or images that “give the illusion of a win when the return is in fact equal to, or below, a stake”.
Stake flexibility and the illusion of control
A product that lets you move from a small stake to a large one in a single tap makes chasing losses easy. A product that lets you press a button to “stop” a spin, or pick your own numbers, can feel like a game of skill when the outcome was already settled. The Commission’s 2021 rules also prohibited features that speed up play or give an illusion of control over the outcome, and banned autoplay on online slots, which it linked to players losing track of their play. The same package set a minimum spin speed of 2.5 seconds, required operators to show total wins, losses and time played during a session, and permanently banned reverse withdrawals.
Access and payment
A machine you have to travel to is different from one in your pocket. The WHO notes that high-intensity products, “including EGMs, high-speed wagering platforms, and casino games, including online versions, are readily accessible almost anywhere”. How you pay matters too: cash in a slot has a natural limit, a debit card does not. Our guide to responsible gambling covers the deposit limits and blocking software that put those limits back.
What the national data shows
Great Britain
The Gambling Commission published a report on 30 January 2025 called Exploring the relationship between gambling activities and PGSI scores. It used year 1 (2023) data from the Gambling Survey for Great Britain, which is collected by push-to-web self-completion.
This is the most useful kind of study for our question, because it adjusts for how many different activities a person takes part in. People who gamble on many things tend to score higher on the PGSI, so without that adjustment every popular product looks risky. The interesting question is which activities stay linked to high scores once that is accounted for.
In the fully adjusted model, these remained significantly associated with a PGSI score of 8 or more:
- Online fruit or slot style games — odds ratio 3.23 (95% confidence interval 1.94 to 5.37)
- Fruit or slot machines played in person
- Casino games, in person, online, or on machines and terminals
- Scratchcards that are not National Lottery scratchcards
- Online instant wins that are not National Lottery instant wins
- Betting on sports or racing in person
- Betting on non-sports events, in person or online
These did not stay significant once everything else was accounted for: National Lottery scratchcards, charity lottery tickets, football pools, online sports and racing betting, online National Lottery instant wins, and bingo played online or in person.
Buying National Lottery draw tickets went the other way, with an odds ratio of 0.62 (0.39 to 0.98). That does not mean lottery tickets protect anyone. It most likely reflects that a large group of people buy a ticket and gamble on nothing else.
The report’s own discussion section is careful about what this proves. The authors write that “the results presented here are associations” and that “there may be other, unmeasured, factors underlying the observed associations”. The analysis cannot separate products that cause more harm from products that attract people already at higher risk. It is based on a single year of data.
Australia
Australia has studied gaming machines for longer than almost anyone. The Productivity Commission’s 2010 inquiry report, published on 23 June 2010, put it simply: “risks of problem gambling are low for people who only play lotteries and scratchies, but rise steeply with the frequency of gambling on table games, wagering and, especially, gaming machines”. It estimated that around 15 per cent of regular gaming machine players were problem gamblers, and that this group accounted for roughly 40 per cent of gaming machine spending.
More recent state data points the same way. The NSW Gambling Survey 2024 — telephone interviews with a random sample of 10,000 adults between March and May 2024 — found that 7.8% of respondents had experienced harm from their own gambling and 12.7% from someone else’s. It concluded that “EGMs are the source of most of the gambling harm in NSW, followed by wagering products, casino table games, online casinos and online poker”.
Note the survey modes. The New South Wales figures come from telephone interviews; the British figures come from an online self-completion survey. People report more gambling problems when nobody is listening, so figures from the two cannot be lined up side by side. We explain that problem in our global harm statistics guide.
The global picture
The Lancet Public Health Commission on gambling, published on 24 October 2024, estimated that about 448.7 million adults worldwide are affected by some level of risky gambling, and around 80 million adults have gambling disorder or problematic gambling.
Its product-level figures are the starkest in the field. Among people who use online casino or slot products, the Commission put problematic gambling at “15·8% of the adults and 26·4% of the adolescents”. Among people who bet on sports, the figures were “8·9% of the adults and 16·3% of the adolescents”. These are rates among users of those products, not shares of the whole population, and they come from pooled international studies using different screening tools.
The WHO’s fact sheet reaches the same conclusion in one line: “Electronic gambling machines (EGMs) and casino games are often associated with the most risk of harm.”
Why rankings differ between countries
You will see different products at the top of different national lists. There are three ordinary reasons for this.
- Availability. New South Wales has gaming machines in pubs and clubs. Norway removed its machines and rebuilt them with mandatory limits. A product cannot cause harm where it does not exist.
- Survey method. Telephone, face to face and online surveys produce different answers to the same question, and different screening tools set different thresholds.
- Product rules. Two products called “online slots” can be very different machines once stake caps, spin speeds and autoplay rules are applied.
What happens when the design is changed
The strongest argument that design matters is what happens when regulators change it.
The clearest example is Britain’s fixed odds betting terminals. On 17 May 2018 the UK government announced that the maximum stake on these machines would fall from £100 to £2, on the grounds that the machines risked large, fast losses. The new limit took effect on 1 April 2019. The same product, with one number changed, became a much slower way to lose money.
The Gambling Commission’s briefing to licensing authorities at the time also predicted something worth remembering: that operators would steer customers towards other machine categories and towards online play, which the stake cut did not touch. Tightening one product often pushes activity to the next one along. That is a real limit on product regulation, not an argument against it.
How to judge a product for yourself
You do not need a scoring tool. The research boils down to a handful of questions you can ask about anything you are about to play.
- How fast is it? How many rounds could you play in ten minutes?
- Does it stop on its own, or only when you decide to stop?
- How quickly can you increase your stake?
- Can you play it anywhere, at any hour?
- Is there anything between you and your bank account?
- Does it feel like skill when it is not?
A product that is fast, continuous, always available and paid for by card is the high-risk combination that shows up in every framework on this page. If that describes what you play, deposit limits, time limits and blocking software are worth setting up before you need them, and self-exclusion is there if limits are not enough.
Where to get help
If gambling is causing you problems, help is free and you do not need a diagnosis to use it. Our guide on how to get help with gambling walks through the options step by step.
- Great Britain: the National Gambling Helpline, run by GamCare, is on 0808 8020 133 and is open 24 hours a day, every day of the year.
- United States: the National Problem Gambling Helpline is 1-800-MY-RESET (1-800-697-3738), for calls and texts. The older number, 1-800-522-4700, still works.
- Elsewhere: ask your national regulator or health service for the helpline in your country.
If you are in immediate danger, contact your local emergency number.
We do not take money from gambling operators and we do not carry affiliate links. You can read how this site is paid for on our funding and independence page, and how we check facts in our editorial policy.
Last reviewed: 25 September 2026. Sources checked on this date.