Blood and urine samples for determination of manganese (Mn) and iron (Fe) concentrations were collected in a cross-sectional study of 137 currently exposed welders, 137 referents and 34 former welders. Aerosol samples for measurements of personal air exposure to Mn and Fe were also collected. The aerosol samples were assessed for their solubility using a simulated lung lining fluid (Hatch solution). On average 13.8% of the total Mn mass (range 1-49%; N = 237) was soluble (Hatch sol), while only 1.4% (<0.1-10.0%; N = 237) of the total Fe mass was Hatch sol. The welders had statistically significantly higher geometric mean concentrations of Mn in whole blood (B-Mn 12.8 vs. 8.0 μg L (-1)), serum (S-Mn 1.04 vs. 0.77 μg L(-1)) and urine (U-Mn 0.36 vs. 0.07 μg g (-1) cr.) than the referents. Statistically significant univariate correlations were observed between exposure to Hatch sol Mn in the welding aerosol and B-Mn, S-Mn and U-Mn respectively. Pearson's correlation coefficient between mean Hatch sol Mn of two days preceding the collection of biological samples and U-Mn was 0.46 (p < 0.001). The duration of employment as a welder in years was also associated with B-Mn and S-Mn, but not with U-Mn. Statistically significantly higher U-Mn and B-Mn were observed in welders currently exposed to even less than 12 and 6 μg m (-3) Hatchsol Mn, respectively. When using the 95(th) percentile concentration among the referents as a cut-point, 70.0 and 64.5% of the most highly exposed welders exceeded this level with respect to B-Mn and U-Mn. The concentrations of B-Mn, S-Mn and U-Mn were all highly correlated in the welders, but not in the referents.