Loneliness is increasingly acknowledged as important in parkinsonism-related disorders. However, little is known about reciprocal longitudinal relationships between loneliness and parkinsonism. In this study, 3,099 community-based individuals (Mage=78.09, 75.2% women) completed measures of loneliness, isolation, depression and were rated on parkinsonism severity annually with 846 having available postmortem data. Controlling for demographic and social-emotional covariates, mixed multilevel modeling evaluated longitudinal impacts of parkinsonism on loneliness and loneliness on parkinsonism. The loneliness→parkinsonism models explained greater variance, with loneliness associated with worse parkinsonism and greater rates of motor decline. Whereas parkinsonism→loneliness models revealed worse parkinsonism associated with worse loneliness, parkinsonism did not predict rate of loneliness change. Subjects with Lewy Bodies at autopsy had worse loneliness proximate to death and a more pronounced time×loneliness interaction compared to those without. There are bidirectional relationships between loneliness and parkinsonism, with loneliness being the more potent longitudinal predictor.