The South Oaks Gambling Screen (SOGS) is widely used to assess the prevalence of pathological gambling. For a variety of reasons, this instrument may not provide an accurate rate of the prevalence of pathological gambling. In this paper, one source of error in data provided by the SOGS is investigated. It is argued that individuals may not fully understand the meaning of some items, and that clarification of the meaning of misunderstood items may in some cases lead to a changed score on the scale. The present study evaluates respondents' understanding of the SOGS items. The results from three studies are reported, each using a different sample: grade school children, adolescents and adults. It was hypothesised that (1) participants would not understand some items of the SOGS, (2) problem gamblers and probable pathological gamblers would be more inclined to interpret items incorrectly than would non-problem gamblers and, (3) consistent with the first two hypotheses, clarification of items would decrease the number of participants identified as problem gamblers or probable pathological gamblers. The data obtained supported hypotheses 1 and 3. Furthermore, hypothesis 2 was supported for grade school children, but not for adolescents or adults. These results are consistent with recent literature on endorsement and acquiescence phenomena, and have implications for prevalence studies of probable pathological gambling.