Between August 1981 and July 1984, 93 patients with polychronotopic superficial papillary carcinoma (Ta and/or T1), flat carcinoma in situ (Tis), or concomitant superficial papillary and in situ bladder carcinoma were entered into a prospective randomized trial of maintenance v nonmaintenance intravesical bacillus Calmette-Guérin (BCG) therapy. Forty-six patients who received BCG weekly for 6 weeks were compared with 47 patients receiving the six-weekly doses of BCG plus monthly BCG for 2 years. Both groups were evaluated every 3 months by cytology, cystoscopy, and biopsy. A significant reduction in the number of recurrent tumors per patient-month was demonstrated for both groups (P less than .0001); however, the difference in reduction of tumors between the two groups was not significant. Additionally, patients receiving maintenance and nonmaintenance therapy had similar tumor recurrence and progression rates. These results indicate that monthly maintenance BCG does not prevent, delay, or reduce tumor recurrence or progression observed with the 6-week regimen. Maintenance BCG was associated with increased local toxicity, primarily dysuria, frequency, and urgency. Dosage reduction was required in 22 of 47 patients (46.8%). When the data were subjected to multivariate analysis, the presence or absence of tumor following induction BCG and PPD skin test results were found to be significant variables. Controlling for either the presence or absence of tumor following induction BCG, tumor recurrence and progression rates were not significantly different for the two treatment groups. However, the absence of tumor after induction BCG was associated with a longer disease-free duration (P = .00001) and time to progression (P = .095). Patients with a reactive tuberculin skin test before and after induction BCG had significantly less tumor recurrences than patients with different PPD skin tests results (P = .02). Tumor progression was not related to tuberculin skin testing.