Objective Testing the hypothesis of an association between knowledge and sexual risk behaviour (SRB) amongst community-clinic workers in Chile, explained by the confounding effect of self-perceived vulnerability to HIV. Methods A cross-sectional survey was analyzed; it was nested within a quasiexperimental study of 720 community-clinic workers in Santiago. The SRB score combined the number of sexual partners and condom use (coded as "high"/"low" SRB). Knowledge of HIV (a 25-item index) was coded as "inadequate"/"adequate" knowledge. Self-perceived vulnerability to HIV was categorised as being "high"/ "moderate"/"low". Control variables included socio-demographics, religiousness and educational level. Percentages/averages, Chi-square tests and logistic regression (OR-estimations) were used for descriptive, association and confounding analysis. Results Respondents were 78.2 % female, 46.8 % married and 67.6 % Catholic. Mean age was 38.9 (10.5 SD) and 69 % had university/diploma level. Self-perceived HIV vulnerability was "low" in 71.5 % cases. A negative association between knowledge and SRB was found (OR=0.55;CI=0.35-0.86), but self-perceived vulnerability did not have a confounding effect on this relationship. This relationship also persisted after being adjusted for multiple control variables (e.g. age, sex, type of primary centre, educational level, and religiousness). Conclusions Some community-clinic workers had inaccurate knowledge of HIV, which was associated with SRB. Self-perceived vulnerability did not have a confounding effect; however, future studies should further analyze occupational risk of HIV as a possible driving factor in health workers´ perception of their risk.Rev. salud pública. 12 (5): [777][778][779][780][781][782][783][784][785][786][787][788][789] 2010