Wideband middle ear power (WMEP) measurement is a method of middle ear analysis that may provide improved diagnostic capability over single-frequency tympanometry. However, normative data, information about test-retest reliability, and results in clinical disorders are needed for clinical application. Normative and reliability data on WMEP in children three days to 47 months of age were obtained using a prototype commercial instrument. A prospective study was conducted in children enrolled from a well-child pediatric clinic (n = 97), with comparisons of age, gender, and middle ear status and stimulus type (broadband chirp and sine wave). No significant age effect for power reflectance across the age range of this study was found, except at 6000 Hz. Significantly higher-power reflectance was found for ears with poor ear status, specifically otitis media with effusion. Smaller but nonsignificant differences in power reflectance were found for ears with positive and negative tympanometric peak pressure. Intraclass correlation coefficients showed significant correlations of 0.68 to 0.97 at various test frequencies using the chirp stimulus. Multivariate analysis of variance showed no significant effect of stimulus type (sine wave vs broadband chirp), ear, or gender. These results provide normative data for wideband middle ear power analysis for infants and children from birth to age four years.