Previous studies have shown a high within-subject correspondence between distortion product otoacoustic emission (DPOAE) input/output (I/O) curves and behaviorally inferred basilar membrane (BM) I/O curves for frequencies above approximately 2 kHz. For lower frequencies, DPOAE I/O curves contained notches and plateaus that did not have a counterpart in corresponding behavioral curves. It was hypothesized that this might improve by using individualized optimal DPOAE primary levels. Here, data from previous studies are re-analyzed to test this hypothesis by comparing behaviorally inferred BM I/O curves and DPOAE I/O curves measured with well-established group-average primary levels and two individualized primary level rules: one optimized to maximize DPOAE levels and one intended for primaries to evoke comparable BM responses at the f(2) cochlear region. Test frequencies were 0.5, 1, and 4 kHz. Behavioral I/O curves were obtained from temporal (forward) masking curves. Results showed high within-subject correspondence between behavioral and DPOAE I/O curves at 4 kHz only, regardless of the primary level rule. Plateaus and notches were equally common in low-frequency DPOAE I/O curves for individualized and group-average DPOAE primary levels at 0.5 and 1 kHz. Results are discussed in terms of the adequacy of DPOAE I/O curves for inferring individual cochlear nonlinearity characteristics.