Background Occupational work involves many factors capable of protecting cognition. The ‘disuse’ hypothesis suggests that removal of such factors at retirement may increase the risk of cognitive decline.

Objective To examine whether retirement is significantly associated with cognitive change after adjusting for preretirement cognitive function, personal, social, health and lifestyle factors, work characteristics and leisure activity.

Methods Participants were from the Whitehall II study, a prospective study of London-based Civil Servants. Short-term memory, the AH4 Part 1 (a test of inductive reasoning), verbal fluency and the Mill Hill Vocabulary Scale were collected at ages 38–60 years, and again, on average 5 years later, at 42–67 years, providing pre- and postretirement cognitive functioning assessments for 2031 participants (470 retired and 1561 working). Linear regression was used to test the association between retirement and cognitive performance adjusted for preretirement cognition.

Results Mean cognitive test scores increased between the two assessments. However, after adjusting for age, sex, education, occupational social class, Mill Hill score, work characteristics, leisure activities, and indicators of physical and mental health, those retired showed a trend towards smaller test score increases over 5 years than those still working, although this only reached 5% significance in one test (AH4; β=−0.7, 95% CI −1.2 to −0.09) and did not show a dose–response effect with respect to length of time in retirement.

Conclusions This trend is consistent with the disuse hypothesis but requires independent replication before it can be accepted as supportive in this respect.