The in vivo effects of the oxazaphosphorine compound ifosfamide (IFO) on human tumour xenografts were assessed in thymus aplastic nude mice. The human origin of the tumours was confirmed by isoenzymatic and immunohistochemical methods. Tumour models were selected from a panel of 180 regularly growing, well-characterized xenografts. The maximum tolerated dose in tumour-bearing nude mice was determined to be 130 mg/kg per day given on days 1-3 and 15-17. After 21 days, lethality was 14% after i.p. and 6% after s.c. administration. A total of 43 human tumours were tested for antineoplastic activity, 15 of which (36%) showed regression: 4/5 breast cancer xenografts, 1/3 colon, 1/1 gastric, 2/7 non-small-cell lung cancers (NSCLC), 3/4 small-cell lung cancers (SCLC), 1/2 sarcomas and 3/3 testicular cancers. Two ovarian, two uterine and six renal cancer xenografts as well as three melanomas and five tumours of various histologies were resistant. In 30 human tumour xenografts, the antineoplastic efficacy of the two oxazaphosphorine derivatives cyclophosphamide and IFO was compared. The maximum tolerated dose of cyclophosphamide was 200 mg/kg per day given i.p. on days 1 and 15; it led to 17% lethality after 21 days. Cyclophosphamide induced tumour regression or remission in 10/30 xenografts (33%) and IFO in 13/30 (43%). In conclusion, the observed efficacy of IFO parallels the clinical situation. Breast, lung and testicular cancer and sarcomas proved to be responsive. The antitumoural activity of IFO shows similarities to that of cyclophosphamide; however, a higher response rate and lower toxicity were noted for the former. Preclinical phase II studies in nude mice seem to offer an effective way of identifying active drugs as well as sensitive tumour types for further clinical development.