Serum microRNAs (miRNAs) have been proposed as novel non-invasive biomarkers for the early detection of cancer. Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) is the most commonly used method for investigating miRNA expression levels, however, the interpretation of RT-qPCR results depends largely on normalization to an appropriate endogenous control. The present study involved 129 patients with non-muscle-invasive bladder cancer (NMIBC), 121 patients with muscle-invasive bladder cancer (MIBC) and 158 healthy controls. The aim of the present study was to determine the most stable reference genes for the investigations of serum miRNA in bladder cancer (BC). MiSeq sequencing was performed and the expression levels of 10 miRNAs and U6 were then measured using RT-qPCR. Following RT‑qPCR, five genes (hsa-miR-193a-5p, hsa-miR-16-5p, U6, hsa-miR-191-5p and hsa-let-7d-3p) were selected for stability analysis using geNorm and NormFinder software. These algorithms identified hsa-miR-193a-5p and hsa-miR-16-5p as the most stably expressed reference genes. The availability of hsa-miR-193a-5p and hsa-miR-16-5p was confirmed in an additional cohort. One-way analysis of variance indicated that no significant differences were present in the expression levels among the three groups. Furthermore, miR-148b-3p was selected as a target miRNA to determine the effect of hsa-miR-193a-5p and hsa-miR-16-5p on miRNA quantification. The combined use of hsa-miR-193a-5p and hsa-miR-16-5p enabled the detection of a significant upregulation of miR-148b-3p in the BC serum. The results of the present study demonstrated that normalization of miRNA data, using a combination of hsa-miR-193a-5p and hsa-miR-16-5p as reference genes, may produce reliable and accurate results for the detection of serum miRNAs in BC.