Magnetopneumography (MPG) is a non-invasive technique to measure the amount of magnetizable dust retained in the lung. This study concerns the relation between MPG measurements of welding fume dust retention in the human lung on the one hand and chest radiography and crude counting of exposure time on the other. Precision and accuracy of the MPG instrument was found to be high. A weak but statistically significant correlation (r = 0.45, P < 0.05) was found between exposure time and retention of magnetizable material in a group of 112 full time manual metal arc welders. The variability was high. Retention rate was estimated on a group level. Chest radiographs of 23 arc welders were reviewed separately in random order by two independent readers who were not aware of MPG findings, occupational anamnesis or the subject's identity and age. No statistically significant correlation was found between MPG and the observer's grading. A weak but statistically significant correlation was found between the readers. We conclude that MPG is highly sensitive to retention of magnetizable material retained in the lung and that the correlation between radiography and MPG reported by previous workers could not be verified in this study. The divergent results may well be due to methodological differences. Investigations using more specific roentgenological methods, such as high resolution computed tomography (HRCT), should be performed.