Objectives: To document: (1) the prevalence of selected risk behaviours of high-school students in the Cape Peninsula; (2) whether the notion of a syndrome of adolescent risk behaviour is valid for this setting, and to investigate whether suicidal behaviour and behaviour exposing oneself to injury should be included in this syndrome; and (3) the relationships among risk behaviours, taking into account their influence upon one another. Design: The study was based on a cross-sectional survey utilising a self-completed questionnaire. Sample: A stratified random sample of 7 340 students was selected from 16 schools in the three major education departments. Results: High prevalences of risk behaviour were recorded; for example, 7,8% had tried to commit suicide in the previous year; 18, 1 % smoke cigarettes; 15,4% had engaged in binge drinking in the previous fortnight; 7,5% had ever smoked cannabis; 37,3% had failed to wear a seat belt on the last occasion this was possible; 9,8% of males had carried a knife to school and 25,0% had walked home alone at night in the previous month; and 17,4% had ever participated in sexual intercourse. With few exceptions, the unadjusted odds ratios for the relationships between pairs of these variables were significant. Between three and nine of 26 risk behaviours qualified for inclusion (p < 0,01) in each stepwise logistic regression model for each gender with each of the above risk behaviours as dependent variables. Cannabis smoking, alcohol bingeing, and exposure to danger in getting home at night were independently associated with having had sexual intercourse. Conclusions: Intervention programmes are needed to reduce the prevalence of risk behaviours in this population. The notion of a syndrome of adolescent risk behaviour is valid for this population, and both suicidal behaviour and behaviour exposing oneself to injury should be included in this syndrome. There are significant relationships between many adolescent risk behaviours even when the influence of other risk behaviours is taken into account. The probability of adverse sequalae of risk behaviours such as exposure to danger in getting home at night and sexual intercourse is amplified by the presence of selected other risk behaviours.