The objective was to determine whether baseline fatty acid intake and erythrocyte omega-3 and omega-6 polyunsaturated fatty acids (PUFAs) can predict risk of total hip arthroplasty (THA) and total knee arthroplasty (TKA) in older women. This was a prospective analysis of 34,990 women in the Women's Health Initiative. Dietary fatty acids were estimated from food frequency questionnaires. Imputed erythrocyte PUFAs were available in a subcohort of 3,428 women. Arthroplasty (THA and TKA), used as a surrogate of severe osteoarthritis, was identified via linked Medicare data. Cox proportional hazards models were constructed to estimate risk of arthroplasty. Risk of THA was associated with higher intake of arachidonic acid, (multivariable hazard ratio [HR] quartile 4 [Q4] vs Q1: 1.16; 95% confidence interval [CI] 1.01-1.34; P = 0.03) and higher intake of eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA; HR Q4 vs Q1: 1.20; 95% CI 1.05-1.39; P = 0.003). There was a linear trend (P = 0.04) for patients to have a higher risk of THA with higher erythrocyte EPA and DHA in body mass index-adjusted models; however, there was no significant difference in patients who had THAs by quartiles of erythrocyte EPA and DHA (P = 0.10). Dietary fatty acids and erythrocyte PUFAs were not significantly associated with risk of TKA. Higher baseline intakes of arachidonic acid and EPA and DHA were associated with a modestly higher risk of THA. No association was found between fatty acids and patients who had TKAs. Further research in populations with direct measures of osteoarthritis severity is needed to better understand the importance of PUFAs in modulating osteoarthritis and arthroplasty risk.