The analysis of protein adducts is a valuable tool for the biomonitoring of humans exposed to alkylating compounds. In case of single or intermittent exposures, however, the significance and validity of adduct analysis is limited due to the typically small increase in adduct levels and a nonlinear adduct elimination kinetic. This issue was addressed in a follow-up observation of four workers accidentally exposed to acrylonitrile (ACN) in a train depot. N-2-Cyanoethylvaline (CEV) in hemoglobin was analyzed in blood samples of the workers and of seven rescue team and hospital members, approximately, 25 days after the accident. Of the 11 potentially exposed persons, only the cleaning workers revealed adduct levels significantly above the general background. Further blood samples of the workers were analyzed after 85, 115 and 175 days, respectively. In these cases, the adduct concentrations declined to background levels associated with individual smoking habits. A linear regression analysis of the data points to a total elimination interval of 148 days, 2 weeks longer than the standard lifespan of erythrocytes and possibly due to a period of acrylonitrile circulation in the blood stream or release from unstable intermediates. The data allow a rough estimate of the initial protein adduct concentration and an evaluation of the internal dose on the basis of biological exposure equivalents. In conclusion, adduct analyses offer valuable information even in the case of short-term exposures.