DPOAEs are thought to be produced by OHC. However, comparison with noise-induced pathology suggests that their origin is much more complex. DPOAE levels (f1&f2 at 55-75dB, f2=1.22 f1, 6 points/octave, 0.5-20kHz) & ABR thresholds were determined preexposure. Noise was a 4kHz OBN at 108dB (1.75h, n=6) or 85dB (24h, n=5). DPOAE level shifts (LS) & ABR threshold shifts (TS) were determined at 0 & up to 30 days post-exposure. The cochleae were fixed with OsO4, embedded in plastic & dissected into flat preparations. The length of the organ of Corti (OC) was measured; missing IHC & OHC counted; stereocilia damage graded; & regions of OC, nerve fiber & stria loss determined. Cytocochleograms were made showing loss/damage by % distance from the apex (& frequency) with the LS & TS overlaid. The best correlation of LS with pathology required plotting at f1. The best correlation of TS was with IHC & nerve fiber loss. Wide regions of up to 40% scattered apical OHC loss showed little LS. In 2 cases, LS occurred with OHC abnormalities but not loss. In 3 cases, there were R/L LS asymmetries with symmetric pathology; in 2 cases, LSs were symmetric with asymmetric pathology. At all recovery times, the largest LS occurred at 3f1-2f2 for mid-frequencies (MF, 4-12kHz) & at 2f1-f2 below & above that. With 108 dB, there was up to 40% DPOAE recovery at MF in 3/6 cases where there was 80-100% OHC loss in the basal half of the OC. Partial recovery at MF also occurred in regions where the OC was missing. With 85 dB, there was no LS at small focal lesions (100% loss of OHC over 0.4 mm) when f1 or f2 was within the lesion but not both. There was no correlation of LS with stereocilia damage. In 2/5 cases with a stria lesion, the only pathological correlate with LS was this damage. These results suggest that either noise-induced DPOAE LSs at MF include a component from the basilar membrane or they are augmented from someplace other than f1 or f2, possibly the basal 20%.