We ascertained nonsteroidal anti-inflammatory drug (NSAID) use in 2,651 participants in the UK MRC treatment trial of hypertension in older adults and measured change in cognitive function over the subsequent 54 months. There was a significant, although modest, association between change in the Paired Associate Learning Test score over time and NSAID use, which was modified by age. NSAID users showed less decline, with younger subjects seeming to benefit more than older. We found no relationship between NSAID use and time taken to complete the Trail Making Test and also no relationship between anti-indigestion drug use and either cognitive outcome. These analyses highlight the need for larger studies with prospective classification of NSAID use and adequate control of confounding, including exposure to other medications. A randomized controlled trial of NSAIDs, in those known to be at risk of cognitive decline or dementia, may be indicated in the future.