Purpose <p>Although human temporal bones suggest that cochlear synapse numbers decline with age and noise exposure, no validated diagnostic method exists. In animal models, cochlear synaptopathy is associated with reduced auditory brainstem response (ABR) wave 1 amplitude and envelope following response (EFR) magnitude for a sinusoidally amplitude modulated (SAM) tone. However, measuring SAM EFR at the optimal modulation frequency (1000&#xa0;Hz) is difficult in humans. A rectangular amplitude modulated (RAM) tone may be more sensitive to synaptopathy, but this has not been validated in animals. In addition, because synaptopathy likely co-occurs with outer hair cell dysfunction (OHC), a diagnostic assay needs to be robust to abnormal auditory thresholds. The objective of this study was to evaluate the relative ability of ABR and EFR measures to predict synapse numbers in mice with varying degrees of synaptopathy and OHC dysfunction.</p> Methods <p>Distortion product otoacoustic emissions (DPOAEs), ABR, SAM EFR, and RAM EFR were recorded from 57 mice with a range of auditory thresholds and degrees of synaptopathy. Cross-validation was used to compare the relative ability of linear regression models incorporating different measures to predict synapse numbers. Predictions were confirmed histologically.</p> Results <p>RAM EFR modulated at 1000&#xa0;Hz was the single best predictor of synapse numbers for broad synapse loss across frequency, while combining RAM EFR with ABR further improved predictions. In contrast, ABR best predicted focal synaptopathy. Incorporating DPOAEs improved predictions for EFR, but not ABR.</p> Conclusion <p>RAM EFR, ABR, and DPOAEs should be used in the future when predicting synapse numbers.</p>