One hundred two patients with recurrent and/or metastatic head and neck squamous cell cancer were entered into four consecutive phase II trials, all cisplatinum (C-DDP, 100 mg/m2/cycle)-based. The two combinations tried were C-DDP, bleomycin, and fluorouracil (CFB) on 54 patients, and cisplatinum and vindesin in 36 patients (CV). The CFB combination was given with C-DDP by continuous infusion over 96 hours (23 patients) or on day 1 (31 patients). The CV regimen was also given in two different schedules, with VDS at 3 mg/m2/g weekly (12 patients) or by a 96-hour continuous infusion (0.6 to 1.0 mg/m2/d) in 24 patients. The following variables: sex, age, performance status, previous therapy, local recurrence, length of disease-free interval (DFI), distant metastases, weight loss, primary site, histological differentiation, type of chemotherapy, previous chemotherapy, evaluable/measurable disease, erythrosedimentation rate, and their relation with response to chemotherapy (WHO) and survival were submitted to both univariate and multivariate analysis (Cox). Overall response rate (RR:CR + PR) was 25 (28%) of 90. In the CFB protocols, RR was 12 (22%) of 54 vs. 13 (38%) of 36 (P = 0.15, NS) in the CV combination group. For the four different combinations the RR was CFB C-DDPci 7 (30%) of 23, CFB C-DDP 1 hour 5 (16%) of 31, CV VDS weekly 2 (17%) of 12, CV VDSci 11 (45%) of 24. The patient populations were very different, with the latest combination consisting of metastatic patients exclusively. Univariate analysis of multiple variables showed age less than 60 years, PS:0 or 1, no previous therapy, absence of local relapse, metastatic disease, long DFI, and that measurable disease was significant for the probability of response. Median survival was 7 months for the 90 evaluated patients, 5 months for nonresponders, and 9 months for responders (P = 0.01). In the univariate analysis, significant factors for survival were PS:0 or 1, a weight loss below 10%, long DFI, response to chemotherapy, erythrosedimentation rate (ESR) of less than 30 mm/1st hr, presence of bone metastasis, and the number of metastases. Multivariate analysis shows PS, the absence of local relapse, and disease-free interval as significant prognostic factors for response. Multivariate analysis factors of significance for survival were PS, weight loss, and response to chemotherapy. The analysis of the clinical pattern showed an evolution in RR from 3 (8%) of 36 on previously irradiated local recurrent disease to 8 (73%) of 11 in previously untreated patients with metastatic disease at presentation.(ABSTRACT TRUNCATED AT 400 WORDS)