What the Data Says About Gambling in Ireland: Participation, Spend and Harm

23 July 2026
By Roger Kennedy
roger@TheCork.ie

What the Data Says About Gambling in Ireland: Participation, Spend and Harm

Numbers about gambling circulate easily in Ireland. One in thirty adults. Five and a half billion euro a year. Half of all the money coming from people already in trouble. They turn up on the radio, in council chambers, in Dail debates and in conversation, usually without anyone saying where they came from or when they were collected.

The odd thing is how thin the underlying evidence base actually is. For a country that gambles as much as this one does, Ireland has produced very few general population studies of gambling. Almost every figure repeated in public traces back to one of two pieces of research, and the two of them disagree with each other by a factor of ten on the single most important question.

Comparing the commercial side of this, the operators, the offers and the licence positions, is a separate exercise again, and readers who want that detail can find a data-led irish casino guide maintained by Tech-Insider, which tracks the Irish online market and how it is regulated. This article does something narrower: it reads the Irish research itself, shows what one survey genuinely establishes, and is honest about the parts it cannot settle.

Ireland has two gambling surveys, not twenty

The first is the Health Research Board’s work. Gambling questions were added to the 2019 to 2020 National Drug and Alcohol Survey at the request of the Department of Justice, and the results were published by the HRB in February 2022 under the authorship of Deirdre Mongan and colleagues. The survey covered 5,762 people aged 15 and over, and it was carried out through interviews rather than online.

The second is the Economic and Social Research Institute’s study, published on 5 October 2023 and written by Diarmaid O Ceallaigh, Shane Timmons, Deirdre Robertson and Pete Lunn. Fieldwork ran in August 2023. It used quota sampling across three separate online panels and produced a usable sample of 2,850 adults aged 18 and over.

Those two designs are not variations on a theme. One asks about the previous twelve months, one about the previous four weeks. One starts at fifteen, one at eighteen. One puts an interviewer in the room, one does not. Almost every apparent contradiction in Irish gambling statistics comes back to those three differences.

How many adults gamble, and on what

The ESRI survey found that 74 per cent of adults reported spending money on some form of gambling in the previous four weeks, and 35 per cent reported gambling online. Lotteries and scratch cards were the most common activities by a distance, followed by betting on horses, dogs and other sports. Casino games and spread betting were the least common.

The HRB survey found that 49 per cent of people aged 15 and over had gambled in the previous twelve months, down from 65 per cent in its 2014 to 2015 predecessor.

Read carelessly, those look irreconcilable: how can four-week participation be higher than twelve-month participation? The answer is that they are measuring different populations by different means four years apart, and the mode of interview matters enormously for a behaviour people are reluctant to discuss. Neither figure is wrong. They are answers to different questions.

The more useful takeaway is about product mix. Whatever the headline participation rate, the median Irish gambler is buying a lottery ticket, not playing roulette. Casino play is a minority activity carried out intensively by a small group, which is exactly why national averages describe hardly anybody.

Case study: reading the ESRI’s 2023 survey line by line

Take one dataset and work through it properly rather than quoting its press release. The ESRI report is the right candidate because it is recent, it is Irish, it is freely available in full, and its authors set out their own doubts at length.

The study’s central design choice was anonymity. International research had found that people under-report gambling to an interviewer, so the ESRI team moved the questionnaire online in the hope that people would answer more honestly without a stranger watching. Responses were then reweighted to Census 2022 figures for age, gender and educational attainment.

On that basis the study estimated that 3.3 per cent of Irish adults meet the threshold for problem gambling, with a 95 per cent confidence interval of 2.5 to 4.0 per cent. That is the one in thirty figure, roughly 130,000 people. Beneath it sat a further 7.1 per cent showing moderate evidence of problem gambling, about 279,000 people, and another 15 per cent, around 590,000, reporting at least one negative gambling experience or behaviour.

Two details rarely survive into the retelling. The first is that confidence interval: the honest reading is somewhere between one in forty and one in twenty-five, not a precise one in thirty. The second is that the three tiers matter more than the headline. Problem gambling in this data is not a wall with people on either side of it. It is a slope.

What that study cannot tell you

The authors are direct about the limits, and repeating them is not an attack on the work.

The sample was drawn from online panels, so it cannot speak for people without internet access, people over the age of eighty, or anyone unable to complete a survey in English. Quota sampling is not random probability sampling, and the report notes that the three panels produced different estimates, which is why the headline figure treats all three equally rather than favouring the ones the authors thought more accurate.

It is a single snapshot taken in August 2023. It cannot show a trend, and because it is cross-sectional it cannot establish cause and effect. Whether problem gambling causes poor mental health, follows from it, or shares a common origin is not something one survey wave can answer, and the report says so.

The spend question carries its own measurement error. The authors flag that some respondents may have interpreted a question about gambling spend as asking about net losses rather than gross stakes. Some of the subgroup breakdowns rest on small numbers: the report notes that its own estimates of spending by men and women within the problem gambling group are likely to be imprecise for exactly that reason.

None of this makes the study unusable. It makes it a study, with a stated design, a stated sample and stated weaknesses, which is more than can be said for most numbers that get quoted in this area.

The spending is nothing like evenly spread

This is where the data becomes genuinely uncomfortable. Mean gambling spend across all adults was 27 euro a week. The median was 7 euro. When an average is nearly four times the midpoint, the distribution has a long tail, and the tail is the story.

The 3.3 per cent classed as problem gambling accounted for 28.3 per cent of all gambling spend. Add the group showing moderate evidence and roughly a tenth of the adult population accounted for 47 per cent of the money. Within the problem gambling group, mean spend was 231 euro a week, about 12,000 euro a year, with a median of 108 euro a week. Slightly more than half of that went online.

The concentration holds even without the harm classification. Among people who had gambled in the previous four weeks, the top quarter by spend accounted for 79 per cent of the total, with a mean of 116 euro a week. The bottom half of gamblers accounted for 8 per cent between them.

It also holds, more weakly, for the products people think of as harmless. Lotteries and scratch cards are the softest end of the market, run largely for good causes, and yet 17 per cent of that spending still came from people with problem gambling, rising to 29 per cent when the moderate group is added.

Anyone who has sat in a money advice office recognises the shape of that distribution. Cork’s MABS branch reported a ten per cent rise in people seeking help during 2025, and household budget problems rarely arrive one at a time. A commercial model in which a tenth of customers supply nearly half the revenue is a model with a small number of very heavy accounts in it, wherever those accounts happen to live.

Why self-reported spending understates real losses

Three separate things push reported spend below actual spend, and the ESRI report shows all three.

Definition is the first. People asked what they spend on gambling often answer with what they think they are down overall, netting off wins they have already restaked. Gross stakes and net losses are different quantities, and the gap between them widens the more often you play.

Recall is the second. Recalling four weeks of small, frequent transactions is genuinely difficult, and frequency in this market is high. Almost a quarter of people who spent money on online slots in the previous four weeks did so at least four or five times a week, and among the problem gambling group that rose to nearly half.

Concealment is the third and the most interesting, because the ESRI tested it directly. Alongside the standard questions, the researchers ran a crosswise experiment, a technique that lets someone answer a sensitive question without the researcher being able to tell what their individual answer was. When one problem gambling question was asked with that additional layer of anonymity, positive responses rose significantly. In other words, moving the survey online reduced concealment but did not eliminate it, and the report concludes that its own estimates may still be too low.

The aggregate check is the cleanest evidence of all. Multiplying mean weekly spend across the adult population implies about 5.5 billion euro a year, against a conservative estimate of 6 to 8 billion euro in gross industry revenue. A survey that was over-recording gambling would land above the industry’s own income. This one lands below it. The money is being spent. It is simply not all being reported.

What the harm measure is actually asking

Both Irish surveys use the Problem Gambling Severity Index, a nine-item scale that asks about consequences rather than amounts: betting more than you could afford to lose, needing to gamble larger sums for the same feeling, borrowing money or selling something to gamble, feeling guilty, being criticised by others, and gambling causing financial or health problems.

That design has a consequence worth understanding. Someone can stake a large sum every week, feel fine about it, cause no household difficulty and score zero. Someone else can stake far less and score highly because the money was borrowed and the guilt is real. The PGSI measures damage, not turnover, and the two are related but not the same.

It also has a blind spot the ESRI picked up. Asked to estimate how many other adults spend more on gambling than they do, respondents systematically overestimated, and people with problem gambling did so most. The internal reference point is broken: heavy gamblers tend to believe their own level of play is ordinary.

Why the two Irish surveys disagree by a factor of ten

The HRB survey put problem gambling at 0.3 per cent of adults, about 12,000 people, with a further 0.9 per cent at moderate risk and 2.3 per cent at low risk. The ESRI put it at 3.3 per cent. Both used the PGSI. Both were competently run.

The likeliest explanation is the interview mode. An in-person survey about a stigmatised behaviour will collect fewer admissions than an anonymous online one, and the ESRI’s own crosswise result shows the effect is real and measurable. Sample selection runs the other way: people with heavy gambling involvement may be more likely to join online panels in the first place, which would push the online estimate up.

The ESRI addressed that with arithmetic rather than assertion. If the true rate were as low as the earlier survey suggested, and if problem gamblers really do supply the share of spend recorded, then ordinary gamblers would have to be spending an average of 41 euro a week to account for industry revenue, against the 16 euro a week they reported. That is a large gap to explain away. The revenue has to come from somewhere.

The honest summary is that the true figure is probably much closer to the higher estimate, that the difference is mostly a measurement artefact rather than a change in behaviour between 2019 and 2023, and that no one can currently say by how much.

What the Irish evidence base still lacks

Ireland has no routine, published national account of gambling participation and spend. There is no annual series, nothing equivalent to the quarterly economic statistics the Central Statistics Office produces, and no public dataset of actual account-level play. Every prevalence figure in circulation is a one-off study, and the two main ones are now several years old.

That may improve. The Gambling Regulation Act 2024 established the Gambling Regulatory Authority of Ireland and provides for a Social Impact Fund financed by operator contributions, along with a national exclusion register. Licensing is being commenced in phases rather than all at once, with betting handled first and remote gaming later in the sequence, so anyone checking the status of a particular sector should check the Authority’s own published position rather than assume. If the regulator eventually publishes licensing and activity data, Ireland will have something closer to a continuous measure instead of occasional snapshots.

Until then, the sensible approach is to read the sources rather than the summaries. The ESRI’s full report, including every table quoted here and its own account of its limitations, is available in full on the institute’s publications site. It runs to just over a hundred pages and is written in plain enough English to be read by anyone who wants to check a figure before repeating it.

How to read the next gambling statistic you see

Six questions handle most of it. Who collected the data, and when. How many people were asked. Were they interviewed in person or answering anonymously. Is the reference period a month, a year or a lifetime. Is the number a stake, a loss or an operator’s revenue, because those differ by a wide margin. And what is the denominator, since a percentage of all adults, a percentage of gamblers and a percentage of online gamblers are three very different things.

Apply those to the one in thirty figure and it holds up as a careful, well-documented estimate with a stated confidence interval and stated limits. Apply them to most of what gets said about gambling in Ireland and rather less survives.

If any of this describes someone you know

Gambling in Ireland is legally restricted to adults aged 18 and over, and none of the figures above are a reason for anyone to feel like a statistic. If the spending patterns in this article look familiar, Problem Gambling Ireland and GamblingCare.ie both offer free, confidential support, Gamblers Anonymous runs meetings around the country including in Cork, and the HSE can point people towards local services. MABS deals with the debt side without judgement and without charge.

Frequently asked questions

Is one in thirty adults really a reliable figure for problem gambling in Ireland?

It is the best current Irish estimate, published by the ESRI in October 2023 from a sample of 2,850 adults, and it carries a confidence interval of 2.5 to 4.0 per cent. The authors themselves say it may be an underestimate rather than an overestimate, because their anonymity experiment showed people still concealing some answers online.

Why does an older Irish survey say 0.3 per cent instead?

The Health Research Board’s 2019 to 2020 survey interviewed people face to face, and people are markedly less willing to admit gambling problems to an interviewer than to an anonymous screen. Sample composition and the four-year gap play some part too, but the interview mode is the leading explanation and the ESRI’s aggregate spend arithmetic supports the higher figure.

Do a small number of players really account for most gambling revenue?

In the ESRI data, yes. Around a tenth of adults, those with problem gambling or moderate evidence of it, accounted for 47 per cent of all gambling spend, and among people who gambled in the previous four weeks the heaviest quarter accounted for 79 per cent of the total.

Why would anyone under-report what they spend on gambling?

Partly by accident and partly not. Many people answer with net losses rather than gross stakes, frequent small transactions are hard to recall accurately over four weeks, and the ESRI’s crosswise experiment showed that some concealment persists even in an anonymous online survey.

Where can I check these figures myself?

The ESRI report is downloadable free from esri.ie, and the Health Research Board publishes its National Drug and Alcohol Survey findings on hrb.ie. Both give sample sizes, fieldwork dates and methods, which is the minimum any gambling statistic should come with before it is worth quoting.