Welders are exposed to a variety of metal fumes, including manganese, that may elevate the risk for neurological disease. This study examines several large data sets to characterize manganese, iron, and total particulate mass exposures resulting from welding operations. The data sets contained covariates for a variety of exposure modifiers, including the presence of ventilation, the degree of confinement, and the location of the personal sampler (i.e., behind or in front of the welding helmet). The analysis suggests that exposures to manganese are frequently at or above the current ACGIH(R) threshold limit value of 0.2 mg/m(3). In addition, there is evidence that local exhaust ventilation can control the exposures to manganese and total fume but that mechanical ventilation may not. The data suggest that higher exposures are associated with a greater degree of enclosure, particularly when local exhaust ventilation is absent. Samples taken behind the helmet were, in general, lower than those measured outside of it. There were strong correlations among manganese, iron, and total particulate mass exposures, suggesting simple equations to estimate one fume component from any of the others.