Eighty-four patients with advanced cancer refractory to conventional therapeutic modalities were randomly assigned in double-blind fashion to one of three intradermal treatment regimens: "high"-dose methanol extraction residue fraction of Bacillus Calmette-Guérin (MER) (2.0 mg); "low"-dose MER (0.5 mg); or 0.9% NaCl solution placebo. Toxicity, consisting primarily of cutaneous inflammation and ulceration, was limited to patients receiving MER and was most severe with the high-dose regimen. Pretreatment clinical and immunological parameters were comparable between patient groups. Although a significant number of patients had increases in various immune parameters according to the criteria used, there was no appearent advantage to MER given in either dosage schedule compared to placebo. Patient survival was not affected by either MER regimen compared to placebo. This investigation failed to demonstrate any significant clinical or immunological benefit from MER given in two dosage regimens in patients with advanced cancer with the laboratory methodology used and emphasizes the importance of appropriate controls in evaluating immunostimulants in humans.