Objective: Cognitive reserve theory seeks to explain the observed mismatch between the degree of brain pathology and clinical manifestations. Early-life education, midlife social and occupational activities and later-life cognitive and social interactions are associated with a more favourable cognitive trajectory in older people. Previous studies of Parkinson's disease (PD) have suggested a possible role for the effects of cognitive reserve, but further research into different proxies for cognitive reserve and longitudinal studies is required. This study examined the effects of cognitive lifestyle on cross-sectional and longitudinal measures of cognition and dementia severity in people with PD. Methods: Baseline assessments of cognition, and of clinical, social and demographic information, were completed by 525 participants with PD. Cognitive assessments were completed by 323 participants at 4-year follow-up. Cognition was assessed using the measures of global cognition dementia severity. Cross-sectional and longitudinal serial analyses of covariance for cognition and binomial regression for dementia were performed. Results: Higher educational level, socio-economic status and recent social engagement were associated with better cross-sectional global cognition. In those with normal cognition at baseline, higher educational level was associated with better global cognition after 4 years. Increasing age and low levels of a measure of recent social engagement were associated with an increased risk of dementia. Conclusions: Higher cognitive reserve has a beneficial effect on performance on cognitive tests and a limited effect on cognitive decline and dementia risk in PD.