IntroductionThe overall prevalence of gambling problems across prison populations is currently unknown. The objective of the present study was therefore to quantitatively synthetize prevalence estimates of gambling problems in prison populations using a random effects meta-analytic model and to investigate if the estimates were moderated by time frame, cut-off levels, and sample size.MethodsTo be included the studies had to report original data on the prevalence of gambling problems in a prison sample and to be written in a European language, whereas data based on abstracts or qualitative reports were excluded. The search ended on December 1, 2023 and were conducted in Web of Science, PubMed, Cinahl, PsycINFO, Embase, Google Scholar, Grey Literature Report, and GreyNet. Risk of bias was assessed with a standardized 10-item measure for epidemiological studies.ResultsA total of 26 studies comprising 9,491 participants were included. The vast majority of the participants were males. The most commonly used instrument for assessment of gambling problems was the South Oaks Gambling Screen. The pooled random-effects gambling problems prevalence estimate was 30.8% (95% CI = 25.1–37.3). The meta-regression analysis showed that none of the three moderator variables (criteria, timeframe, sample size) were related to the gambling problems prevalence. Common limitations of the included studies entailed not being representative nationally or for the target population, lack of randomization, and low response rate. The meta-analysis was restricted to studies published in a European language.ConclusionsOverall, the studies show that 1 in 3 prisoners has gambling problems and suggests that more emphasis on relevant prevention and treatment is warranted for this population. The study was funded by the Norwegian Competence Center for Gambling and Gaming Research and pre-registered at PROSPERO (CRD42023390552).