Occupational exposure to 2,4- and 2,6-toluene diisocyanate (2,4- and 2,6-TDI) was measured during the production of flexible foam. The usefulness of urinalysis of the TDI-derived amines, 2,4- and 2,6-toluenediamine (2,4- and 2,6-TDA), for exposure assessment was compared with air monitoring. Urine samples were collected from 17 employees at two plants. The workers' personal exposure was measured using 1-(2-methoxyphenyl)-piperazine (2MP)-impregnated glass fibre filters for sampling and high-performance liquid chromatography (HPLC) with ultraviolet (UV) and electrochemical (EC) detection for quantification. The limit of detection (LOD) of 2,4- and 2,6-TDI was 0.01 microtg ml(-1) for a 20 microl injection. The precision of sample preparation, expressed as the relative standard deviation (RSD), was 0.6% with UV detection and 0.8% with EC detection at a 2,4-TDI concentration of 0.2 microg ml(-1) (n = 6). For 2,6-TDI, the corresponding RSDs were 0.5% and 0.8%. The urinary 2,4- and 2,6-TDA metabolites were determined after acid hydrolysis as heptafluorobutyric anhydride derivatives by gas chromatography-mass spectrometry. The LOD in urine was 0.35 nmol l(-1) for 2,4-TDA and 0.04 nmol l(-1) for 2,6-TDA. The precision (RSD) of six analyses of human urine spiked to a concentration of 100 nmol l(-1) was 3.7% for 2,4-TDA and 3.6% for 2,6-TDA. There was a trend for linear correlation between urinary TDA concentration and the product of airborne TDI concentration and sampling time. Urinalysis of TDA is proposed as a practical method for assessing personal exposures in workers exposed intermittently to TDI.