A respiratory sample survey of 609 shipyard workers was conducted in 1979: the men were reassessed an average of 7.2 years later. The 53 deaths between the surveys were related to age, level of lung function and smoking but not to trade as a welder or caulker/burner. Of the survivors, 488 (88%) were seen, including 425 men who had retired or been made redundant. Redundancy was related to age, smoking, and respiratory symptoms; the average reduction in duration of employment per symptom was 0.44 years. Changes in respiratory symptoms included onset of chronic bronchitis and wheeze on most days (numbers respectively 77 and 109) and increased breathlessness on exertion (n = 89); significant related factors included smoking, previous metal fume fever or pneumonia, and, for breathlessness, trade as a welder or caulker/burner. Electrocardiographic evidence for myocardial ischaemia was also associated with increased breathlessness. The annual declines in FEV1 and other spirometric indices were related to age, to being a smoker at the time of the initial survey, and to trade as a welder or caulker/burner compared with trades that did not involve welding or burning. There was significant interaction between these effects. In a subsample of 124 redundant workers there was also significant interaction between the effects of fumes and atopy (skin test positive to common antigens) or a raised serum IgE concentration. It was concluded that welding fumes interacted with smoking and an atopic constitution to cause respiratory impairment. The results related mainly to exposures in the past and were not necessarily relevant for present day conditions in the industry.