OBJECTIVES: A high-strain job (a combination of high job demands and low job control) is expected to increase the risk of health problems, whereas an active job (high demands and high control) can be hypothesised to be associated with a greater capacity to learn. We tested associations between high-strain and active jobs and cognitive function in middle-aged men and women. METHODS: Data on 4146 British civil servants (2989 men and 1157 women) aged 35-55 years at baseline came from the Whitehall II study. Cumulative exposure to both high-strain and active jobs was assessed at phases 1 (1985-1988), 2 (1989-1990) and 3 (1991-1993). Cognitive performance was assessed at phases 5 (1997-1999) and 7 (2003-2004) using the following tests: verbal memory, inductive reasoning (Alice Heim), verbal meaning (Mill Hill), phonemic and semantic fluency. Analyses were adjusted for age, sex and employment grade. RESULTS: Longer exposure to high job strain and shorter exposure to active jobs were associated with lower scores in most of the cognitive performance tests. However, these associations disappeared on adjustment for employment grade. Phonemic fluency was an exception to this pattern. Associations between exposure to an active job and phonemic fluency at both follow-up phases were robust to adjustment for employment grade. However, there was no association between exposure to active jobs and change in phonemic fluency score between the follow-up phases after adjustment for employment grade. CONCLUSIONS: In these data, associations between cumulative exposure to high-strain or active jobs and cognition are largely explained by socioeconomic position.