OBJECTIVES: Attitudes may influence behaviour through both deliberative and automatic processes. To investigate the automatic influences of attitudes, this study explores the accessibility of modally salient beliefs about physical activity outcomes using response latency measures. DESIGN: Response latencies for modally salient beliefs for physically activity outcomes were compared with latencies for non-salient, hygiene outcomes. Possible relationships between self-report and response latency was assessed between- and within-subjects. METHOD: Regularly active participants (N=148) completed a computer-based response latency task in which they indicated whether an outcome, for example more fit, was a likely or unlikely consequence of six different physical activities, for example go running. Self-reports of the likelihood of these outcomes, their importance, intentions to participate in the physical activities and frequency of participation were obtained. RESULTS: As expected, the physical activity outcomes were more accessible than control outcomes. In addition, the outcome strong heart was less accessible than the outcomes more fit and have fun. There was only weak evidence, however, of any relationship between self-reports and the accessibility of the physical activity outcomes. CONCLUSION: Response latency data may represent a source of between-subject variation that differs from self-report. Discussion focuses on the possible origins of such a discrepancy.