Résumé : In this paper, the mechanism of the nitrobenzylpyridine (NBP) method to measure the alkylating activity of drugs originally described by Epstein et al. [J. Epstein, R.W. Rosenthal, R.J. Ess, Anal. Chem. 27 (1955) 1435-1439] and modified later by others was revisited using melphalan, m-sarcolysin, chlorambucil, cyclophosphamide and ifosfamide. Its direct application to determine the activity of these drugs in human serum and aqueous media is described and discussed. This method, based on the formation of a chromophore due to the reaction between the alkylating agent and NBP, was significantly improved by extracting as quickly as possible the reaction product(s) into chloroform before adding alkali to develop the color. This significantly limited the degradation by hydrolysis of the products and enhanced the yield of the end chromophore in the organic phase. The reaction time was optimized by monitoring each compound color development. The best reaction time for each compound was selected and a higher stability of the extracted color over at least 1h was obtained (compared to a couple of minutes in previous studies). Most interestingly, water evaporation due to heating had little or no effect on the linearity of standard curves evaluated in the micromolar concentration range. Both the sensitivity and reproducibility of the method were therefore significantly improved. There appears to be a direct correlation between compound hydrolysis and alkylation activity; the relative reactivity is different among the compounds owing to the rate of (i) production, (ii) the relative proportions and (iii) the hydrolysis of the intermediates. A general mechanism for the nucleophilic competitive substitution is proposed.