Distortion product otoacoustic emission (DPOAE) (2f1-f2) amplitude is dependent upon both the frequency ratio and level separation of the eliciting primary tones. In adults it has been established that, on average, a f2/f1 ratio of 1.22 is optimal for evoking the most robust DPOAE; DPOAE amplitude is systematically reduced when f2/f1 ratio is either increased or decreased, thus forming a bandpass function (Harris et al., 1989). The frequency ratio function (DPOAE amplitude x f2/f1 ratio) is thought to reflect the filtering properties of the cochlea (Allen and Fahey, 1993; Brown et al., 1993). Primary tone level separation also influences DPOAE amplitude, with a 10- to 15-dB level difference between the primary tones (L1 > L2) typically generating largest amplitude in adults. Equivalent studies have not been conducted in neonates. The present study evoked the 2f1-f2 DPOAE in adults, term and premature neonates to define the optimal f2/f1 ratio and L1-L2 level separation and to investigate the filtering properties of the developing cochlea. Two f2 frequencies were investigated: 1500 and 6000 Hz. F2 was held constant while f1 was varied to produce 13 frequency ratios. Primary tone level separation varied from 15 to 0- in 5-dB intervals. ANOVA were conducted on the resulting f2/f1 frequency ratio and level separation data. Results showed that the mean optimal frequency ratio for DPOAE generation is comparable in adults and neonates. Also, either a 15- or 10-dB level separation (L1 > L2) produced the largest DPOAE amplitude for adults and term neonates whereas DPOAEs from premature neonates appeared to be relatively insensitive to primary tone level separation. The f2/f1 frequency ratio functions were similar in shape, slope, and bandwidth for adults and neonates, suggesting adult-like cochlear filtering prior to term birth. This finding is in agreement with previous work from our laboratory reporting adult-like DPOAE suppression tuning curves in term-born neonates [Abdala et al., Hear. Res. (1996)].