The Problem Gambling Severity Index, the scored module of the Canadian Problem Gambling Index, is a population-based survey instrument that is becoming the preferred epidemiological tool for estimating the prevalence of disordered gambling. While some validation evidence for the Problem Gambling Severity Index is available, very little is known about its psychometric characteristics in developing countries or in countries the populations of which are not highly Westernised. The aim of this study was to investigate the validity of the Problem Gambling Severity Index with a specific focus on its criterion-related and construct (concurrent) validity in a community sample of gamblers in South Africa (n = 127). To this end, the Problem Gambling Severity Index was administered alongside the Diagnostic Interview for Gambling Severity and measures known to associate with gambling severity (impulsivity, current debt, social problems, financial loss, race, sex). Results showed that the Problem Gambling Severity Index was predictive of Diagnostic Interview for Gambling Severity diagnosis from both a categorical and dimensional point of view and demonstrated high discrimination accuracy for subjects with problem gambling. Analysis of sensitivity and specificity at different cut-points suggests that a slightly lower Problem Gambling Severity Index score may be used as a screening cut-off for problem gambling among South African gamblers. The Problem Gambling Severity Index also showed significant correlations with the Barratt Impulsiveness Scale, a widely known measure of impulsivity, and with some of the predicted behavioural variables of interest (gambling activities, money lost to gambling, current debt, interpersonal conflict). This article therefore demonstrates initial criterion and concurrent validity for the Problem Gambling Severity Index for use in South African samples.