To assess the influence of foundry exposure on malignant and non-malignant respiratory disease, the proportional mortality ratio (PMR) was used to compare the cause of death distributions of the 578 dead members of the Iron Moulders Society of South Africa, recipients of the union's death benefit fund between 1961 and 1983. Comparisons were made with the age and period specific white male deaths. For the 419 members where job information was available, the influence of occupation (journeyman, production moulder) was assessed using different techniques--the relative proportional mortality ratio (RPMR), the mortality odds ratio (MOR), and the proportional cancer mortality ratio (PCMR) for comparison. Excess PMRs were found for cancer of the trachea, bronchus, and lung (1.71, p = 0.03; Poisson one sided test) for those over 65 and for non-malignant respiratory disease (1.58, p = 0.01) and for injuries and poisonings (2.61, p less than 0.0001) in those under 65. Reduced PMRs were found for all cancers (0.75, p = 0.03) and all circulatory disease (0.91, p = 0.12) in those under 65. When comparing job types, raised risks were obtained for journeymen using all methods (RPMR, MOR, PCMR) but the small cell sizes rendered the results non-significant. The raised PMRs due to respiratory disease are unlikely to be due to smoking because of a poor association with other causes of death related to smoking. A more likely explanation is that these excess rates for malignant and non-malignant respiratory disease are due to exposure to the foundry environment. Of additional concern are the high PMRs due to injuries and poisonings, which could be related to the high accident rates in the iron and steel industry.