The bioavailability of three administration forms of cyclophosphamide (CP) was determined by comparison with the i.v. application in 12 female patients with breast cancer. CP charges were given on four consecutive days at a dosage of 175 mg/m2 (50-mg dragees) in a randomized sequence. CP blood levels were measured with N/P flame ionization gas chromatography. The average half-life of intravenously applicated CP and for all three oral formulations was about 4 h. The ratio AUC p.o./AUC i.v. was 0.896 with the gastric juice-resistant formulation of CP, and 0.914 and 0.958 with the two "rapid release" formulations (soluble in gastric juice). The slight differences in the bioavailability and in the peak concentration (19.1; 22.1; 22.8 nmol/ml) were not significant. However, the gastric juice-resistant formulation displayed delayed peak times (2.5 h as compared to 1.13 and 1.42 h) as well as an irregular absorption phase with several maxima in the blood level curves in a few patients. When a first-pass effect of approximatively 8% is added to the bioavailability of the nonmetabolized CP, an almost complete absorption results for all CP formulations. Since the first-pass effect by hepatic metabolism of CP represents no detoxication, but a conversion of inactive CP into the actual cytostatically active form, a cytostatic bioavailability of likewise almost 100% can also be assumed.