Objectives. We used data from SWEOLD, a Swedish nationally representative study of individuals aged 77 years or older, to examine midlife indicators of job strain in relation to cognitive performance and impairment. Methods. In all, 827 participants completed an abridged 11-point version of the Mini-Mental State Examination in-person in 1992 and/or 2002 and had self-reported and/or occupation-based scores for job control and demands from data collected in 1968. Seventeen percent scored below the cutoff for cognitive impairment. Results. Controlling for age, sex, education, self-rated health, and year of cognitive screening, low self-reported and occupation-based job control at midlife was associated with poorer cognitive performance later (ps < .001). For the occupation- based measure, low job control was also associated with greater likelihood of impairment, whereas having an active job (high job control/high job demands) was associated with better cognitive performance and lower likelihood of impairment (ps < .01). Childhood environment, midlife depressive symptoms, and social activity had limited influence, whereas the influence of both adulthood socioeconomic position and work complexity on these results was more pronounced. Discussion. Job control at midlife, by itself and in combination with job demands, may influence cognitive functioning later above and beyond demographic variables and other occupational characteristics. © The Author 2011.