Human head and neck tumours were successfully transplanted in athymic nude mice. In 14 xenograft lines the effect of 1 to 3 clinically active agents could be tested. Maximum tolerated doses were given daily for 3-7 days. Growth delay was estimated in terms of the number of volume doubling times gained by the treatment. Cis-platinum and bleomycin appeared to be effective agents. In all 6 lines in which cis-platinum was examined, growth delay sometimes followed by complete regression was achieved. In 6/7 lines a response to bleomycin was observed. There was wide variation in sensitivity to cis-platinum and bleomycin among the different lines. Methotrexate, effective in 40-60% of patients with head and neck cancer, essentially showed no activity. Methotrexate produced a minimal growth delay in 1/11 lines treated. Two of the patients from whom xenografts were obtained responded to methotrexate treatment. The observed lack of activity of methotrexate against these tumour xenografts indicates that this model has limitations in the screening of new anticancer agents.