OBJECTIVE To test the hypothesis that cardiovascular reactivity to laboratory mental stressors interacts with job strain in predicting blood pressure at work. METHODS Ambulatory monitoring of blood pressure and heart rate was carried out for an 8-h period on a work day and on an equivalent non-work day in 49 male firefighters. METHODS Participants were recruited from a larger cohort (n = 90) on the basis of showing high or low systolic reactions to mental arithmetic 15-24 months previously, coupled with high or low ratings of perceived job strain (high demand-low control). Four groups were tested: low job strain-low systolic reactors (n = 12), low job strain-high systolic reactors (n = 12), high job strain-low systolic reactors (n = 12) and high job strain-high systolic reactors (n = 13). RESULTS Systolic blood pressure (SBP) was higher on work than non-work days, and diastolic blood pressure and heart rate were higher at work in the morning but not in the afternoon. These effects were due partly to posture and physical activity differences between the two days. Neither job strain nor laboratory reactivity independently predicted ambulatory blood pressure. However, SBP was significantly higher during the afternoon at work in the high job strain-high systolic reactors than in the other groups. This was independent of baseline SBP, and was not due to differences in posture or activity at the time of recordings. Ambulatory SBP reactivity (difference between ambulatory values and workplace resting levels) in the afternoon at work was also elevated significantly in high job strain-high systolic reactors compared with in the other groups. CONCLUSIONS The results support the hypothesis that individual differences in the appraisal of work stress modulate the relationship between stress reactivity and ambulatory blood pressure.