OBJECTIVE: To examine associations of welding and manganese exposure with Parkinson disease (PD) using meta-analyses of data from cohort, case-control, and mortality studies. METHODS: Epidemiologic studies related to welding or manganese exposure and PD were identified in a PubMed search, article references, published reviews, and abstracts. Inclusion criteria were 1) cohort, case-control, or mortality study with relative risk (RR), odds ratio (OR), or mortality OR (MOR) and 95 confidence intervals (95% CI); 2) RR, OR, and MOR matched or adjusted for age and sex; 3) valid study design and analysis. When participants of a study were a subgroup of those in a larger study, only results of the larger study were included to assure independence of datasets. Pooled RR/OR estimates and 95% CIs were obtained using random effects models; heterogeneity of study effects were evaluated using the Q statistic and I(2) index in fixed effect models. RESULTS: Thirteen studies met inclusion criteria for the welding meta-analysis and 3 studies for the manganese exposure meta-analysis. The pooled RR for the association between welding and PD for all study designs was 0.86 (95% CI 0.80-0.92), with absence of between-study heterogeneity (I(2) = 0.0). Effect measures for cohort, case-control, and mortality studies were similar (0.91, 0.82, 0.87). For the association between manganese exposure and PD, the pooled OR was 0.76 (95% CI 0.41-1.42). CONCLUSIONS: Welding and manganese exposure are not associated with increased PD risk. Possible explanations for the inverse association between welding and PD include confounding by smoking, healthy worker effect, and hormesis.