The object of this study was to investigate whether exposure of pipe-layers to thermal degradation products of diphenylmethane diisocyanate (MDI) could be assessed by analysing 4,4-methylenedianiline (MDA) in hydrolysed plasma and urine, and whether the genotype for N-acetylation affected these biomarker levels. Blood and urine samples were drawn from 30-pipe-layers who had been welding polyurethane (PUR) insulated pipes during the preceding 3 months. MDA in hydrolysed plasma and urine was determined with a gas chromatography-mass spectrometry technique, and genotype for N-acetylation was analysed with a polymerase chain reaction technique. MDA in plasma was detected in 18 of the 30 pipe-layers. Their plasma concentrations of MDA varied from 0.05 to 8.48 micrograms/l. There was a significant negative correlation between time since last welding of PUR-insulated pipes and P-MDA (rs = 0.50, P = 0.005). There was also a significant positive correlation between the estimated number of welded PUR-insulated pipes during the preceding 3 months and P-MDA (rs = 0.68, P = < 0.001). No significant association between genotype of N-acetylation and P-MDA was observed in a multiple regression analysis when adjustment was made for the estimated cumulative exposure to thermal degradation products of MDI. MDA in urine was detected in only four of the 30 pipe-layers. These four subjects had been welding PUR pipes on the same day as the sampling, or on the day before. The present results indicate the spot plasma samples analysed for MDA may give a rather good estimate of exposure to MDI during the preceding months. P-MDA, but not U-MDA, therefore seems to be a useful biomarker of long-term exposure to MDI. The individual N-acetylation capacity did not affect the plasma levels of MDA.