The level of N-(2-hydroxypropyl)valine adducts in haemoglobin has been shown to correlate well with workplace exposure to propylene oxide (PO). However, the analytical method, using the modified Edman degradation procedure, is prohibitively time-consuming and expensive for use as a routine workplace exposure measurement tool. As an alternative, AB Biomonitoring Ltd of Cardiff, Wales, developed a competitive immunoassay for the determination of N-(2-hydroxypropyl)valine adducts in human haemoglobin. Studies showed that whole blood samples analysed using an enzyme linked immunosorbent assay (ELISA) and the modified Edman degradation procedure over the concentration range 3.7-992 nmol N-(2-hydroxypropyl)valine g(-1) haemoglobin are in good agreement (correlation coefficient 0.998, n = 10). The intervariance and intravariance data indicate the repeatability of the ELISA method over the assay conditions employed and show that it is robust over its working range [2-200 pmol N-(2-hydroxypropyl)valine g(-1) haemoglobin]. The assay employs a whole blood matrix and has a working range of 2-6000 pmol g(-1) Hb (equivalent to up to 5 ppm PO exposure, 8 h per day, 5 days per week, over 4 months). The practicality of the assay was tested by assessing exposures to PO at three world-scale manufacturing sites in France and The Netherlands. Over 800 samples were taken over a 2 year period from operators, maintenance fitters and office staff. The data, typically <50 pmol g(-1) globin, indicate that exposures were significantly <0.1 ppm at all times (The Dutch occupational exposure limit is 2.5 ppm over 8 h). Samples were taken after a major turnaround and also before and after the start-up of a newly commissioned plant. All data indicated that high levels of control were effective in minimizing exposure. This study has shown that the immunoassay is a powerful tool for the exposure component of future epidemiology studies, as well as a definitive demonstration of the effectiveness of exposure controls.