Background: Large scale gambling research has traditionally used screening instruments such as the Problem Gambling Severity Index (PGSI) to determine prevalence rates and inform policy. Implicit within this approach is the assumption that scores on screening instruments are a proxy for the presence and severity of harms associated with gambling. Method: A sample of 99 clients attending a specialist gambling treatment clinic, and 330 regular gamblers in the community were administered the Problem Gambling Severity Index and an instrument measuring severity of harm and its relationship to gambling. Results: Results suggested that gamblers in the community self-reported experiencing lower overall levels of harm across seven domains including psychological distress compared to the clinical sample. Similar patterns were observed when examining participants meeting PGSI criteria for problem gambling, and for those problem gamblers experiencing any level of harm. Conclusions: Results suggest that high levels of psychological and financial harm are a robust characteristic of individuals in treatment. Although overall rates of harm attributed to gambling were lower in the community, approximately a fifth of the sample failing to meet problem gambling cutoff scores reported harms. Exclusively using screening instruments may results in incorrect estimates of the number of gamblers experiencing harm.