Isosorbide 2,5-dinitrate and its pharmacologically active metabolites, isosorbide 2-nitrate and isosorbide 5-nitrate, in plasma accumulated to the predicted steady-state after five consecutive oral doses of sustained-release tablets containing 40 mg isosorbide dinitrate at 12-h intervals and after five consecutive oral doses of reference standard-release tablets containing 20 mg at 6-h intervals to 12 subjects in a crossover study. The comparative bioavailability of isosorbide dinitrate, isosorbide 2-nitrate and isosorbide 5-nitrate from the sustained-release tablet was 110 per cent (p greater than 0.05), 89 per cent (p greater than 0.05), and 89 per cent (p less than 0.05), respectively, of that from the reference standard-release tablet. The isosorbide dinitrate plasma level data were the more variable, as expected for a drug of low systemic availability subject to extensive first-pass elimination. The posterior probability that the true bioavailability of isosorbide dinitrate was included within the usually accepted limits of 80-120 per cent was 0.74, a value which is probably insufficient to justify claims of bioequivalence with the reference formulation in respect of extent of availability. In contrast, the posterior probability that the bioavailability of the metabolites isosorbide 2- and 5-nitrate was included within these limits was 0.90 and 0.98, respectively. On the basis of the mononitrate data, these two formulations may be judged bioequivalent in respect of extent of availability despite a formal statistically significant formulation-related effect in the analysis of variance of the isosorbide 5-nitrate bioavailability data. Claims of bioequivalence of isosorbide dinitrate sustained-release formulations may be more economically justified by analysis of the mononitrate plasma concentrations, although concentrations of the formulated parent dinitrate should also be known.