The preliminary results of a randomized, multicentre, prospective trial conducted by the Spanish Oncology Group (CUETO) are presented. The primary aim of the trial was to determine whether or not lowering the dose of bacillus Calmette-Guérin (BCG) could reduce toxicity without compromising efficacy. From February 1991 to June 1992, 500 patients with superficial bladder cancer (TaG2-3, T1G1-3) entered the trial. Of these patients, 252 were included in group A (BCG, 81 mg) and 248 in group B (BCG, 27 mg). There were no significant differences in clinical and pathological characteristics between the two groups. Following treatment, significant differences arose between groups A and B in local severe toxicity (22.6 vs. 4.2%, respectively; p < 0.01). Systemic toxicity was more common in group A than in group B (p < 0.01): pulmonary effects (2.3 vs. 0.4%), fever (26.9 vs. 12.9%) and malaise (16.2 vs. 8.4%), all p < 0.01. Regarding efficacy, 204 patients in group A and 210 in group B were evaluated after a mean follow-up of 18.6 months. No significant differences in recurrence rate (18.1 vs. 19.5%) or progression rate (2.4 vs. 4.8%) were found. However, in patients with superficial bladder cancer with high risk of recurrence and progression (carcinoma in situ and G3), caution must be exercised in selecting the dose.