In view of the great psychological effect deafness has on one's social life, the scientific community has long sought the best way to define the hearing function. The development of new technologies in this field has set "early intervention" as the primary target for screening. For this reason, within the contest of a program of audiological newborn hearing screening, the Authors have used data from a selected group of 1250 newborns to (a) compare the clinical feasibility and performance of three different DPOAE protocols and (b) establish the scoring criteria defining "pass" or "fail" responses. All subjects participating in this study were randomly selected and their normal hearing was verified by linear TEOAE recordings. The test was carried out, using an Otodynamics ILO92 Analyser version 5.60, on the second day of life or later (in case of extended recovery), during natural sleep and after feeding. The DPOAE recordings were elicited using asymmetric stimuli with L1 > L2 and a frequency-ratio of 1.22 following three different SPL protocols: 60-50 dB (PR1), 65-55 dB (PR2) and 75-65 dB (PR3). Five frequencies of the cubic distortion product (referring to f2) were tested with an ILO macro at 1.5 KHz, 2.0 KHz, 3.0 KHz, 4.0 KHz and 5.0 KHz. The statistical evaluation of differences in the signal-to-noise ratio (S/N) between the PR1 and PR2 protocols showed no significant differences. On the other hand significant differences were found between the PR3 and the PR2 S/N ratios, the former providing the best performance (higher values). The scoring criteria were defined by minimum free distribution tolerance intervals of the S/N ratios at the five tested frequencies. In conclusion this study confirms the feasibility of DPOAE recordings in a un iversal newborn hearing screening program and shows the appropriate pass-fail parameters to be used for this purpose. It should be noted that this approach enables acquisition of frequency-specific information which might further improve audiological diagnosis.