Welding fumes contain various metals. Past studies, however, mainly focused on Manganese (Mn)-related neurotoxicity. This study investigated welding-related mixed metal exposure effects on MRI metrics in the basal ganglia (BG) and their dose-response relationship. Subjects with (N = 23) and without (N = 24) a welding exposure history were examined. Metal exposure was estimated with exposure history questionnaire and whole blood metal levels. T1 (weighted-intensity and relaxation time; estimates of brain Mn accumulation), diffusion tensor imaging [Axial (AD), mean (MD), radial diffusivity (RD), and fractional anisotropy (FA); estimates of microstructural differences] metrics in BG [caudate nucleus, putamen, and globus pallidus (GP)] and voxel-based morphometry (for volume) were examined and related with metal exposure measures. Compared to controls, welders showed higher GP R1 (1/T1; p = 0.034) but no differences in blood metal and T1-weighted (T1W) values in any ROIs (p's > 0.120). They also had higher AD and MD values in the GP (p's < 0.033) but lower FA values in the putamen (p = 0.039) with no morphologic differences. In welders, higher blood Mn and Vanadium (V) levels predicted higher BG R1 and T1W values (p's < 0.015). There also were significant overall metal mixture effects on GP T1W and R1 values. Moreover, GP AD and MD values showed non-linear associations with BG T1W values: They increased with increasing T1W values only above certain threshold of T1 values. The current findings suggest that Mn and V individually but also metal mixtures jointly predict GP T1 signals that may in turn contribute to altered DTI metrics in the BG after certain exposure threshold levels.