OBJECTIVES: To figure out optimal bacillus Calmette-Guerin (BCG) maintenance schedules for non-muscle-invasive bladder cancer (NMIBC) patients by comparing different schedules in a systematic review using conventional and network meta-analysis. MATERIALS AND METHODS: Literature was searched in the databases of Medline, Embase, Cochrane library, Clinicaltrials.gov, Wanfang, CNKI and SinoMed in April 2016 and 9 randomized clinical trials comparing intravesical BCG maintenance therapy with BCG induction-only therapy or comparing different BCG maintenance schedules (induction-only, 1 year, 1.5 year, 2 year, 3 year maintenance) in NMIBC patients were included. Conventional and network meta-analyses within a Bayesian framework were performed to calculate odds ratios of tumor recurrence, progression and side effects (cystitis, hematuria, general malaise and fever). The surface under the cumulative ranking curve (SUCRA) mean ranking was used to obtain schedule hierarchy. RESULTS: Data from 1951 patients showed that longer-term maintenance BCG therapy does not significantly decrease tumor recurrence and progression rate of NMIBC compared to shorter-term maintenance BCG therapy. However, longer-maintenance therapy does not increase side effect incidence compared to induction-only therapy. According to SUCRA results, induction-only therapy has the highest probability of recurrence and progression but least probability of side effects. CONCLUSIONS: Longer BCG maintenance therapy (such as 3 years) is not superior to shorter maintenance therapy (such as 1 year). But maintenance therapy overall is better than induction-only BCG therapy while not increasing side effects. Though further evidence and clinical practice with balanced confounding factors (risk stratification and BCG strain) are wished for, the current study suggests the common use of 1 year intravesical BCG instillation for NMIBC patients.