CONCLUSIONS The distortion product otoacoustic emission (DPOAE) amplitudes in diabetics were significantly lower than those in controls when pure tone audiometry (PTA) was within normal limits, implying cochlear function impairment in the early stage of diabetes mellitus. Retrocochlear damage was suggested by elongated wave latencies and the wave interval of auditory brainstem response (ABR) in diabetics. OBJECTIVE This study aimed to systematically analyze the application of DPOAE in evaluation of the hearing function of diabetics. METHODS Eligible articles were identified through searches of nine different electronic databases. Two investigators reviewed the original articles independently, with pre-defined inclusion and exclusion criteria. Meta-analyses were conducted by using Metan module. RESULTS There were seven articles eligible for the analysis. PTA thresholds were within normal limits in all diabetics at low-middle frequencies. The mean DPOAE amplitudes of diabetics were significantly lower than those of controls. The standardized mean difference (SMD) (95% CI) was -0.49 (-0.70, -0.27), -0.46 (-0.63, -0.30), and -0.60 (-0.77, -0.43) at 1, 2, and 4 kHz, respectively. The latencies of waves I, III, and V in diabetes were significantly longer than those of controls (0.43 (0.23, 0.64), 0.54 (0.33, 0.75), and 0.56 (0.35, 0.77), respectively). The wave interval I-V was significantly longer in diabetics (p < .001, 0.47 [0.24, 0.70]).