Background: Deficient rapid auditory processing (RAP) has been associated with early language impairment and dyslexia. Using an auditory masking paradigm, children with language disabilities perform selectively worse than controls at detecting a tone in a backward masking (BM) condition (tone followed by white noise) compared to a forward masking (FM) condition (tone preceded by white noise). Tallal's (1980) auditory processing hypothesis posits that abnormal RAP leads to reduced (or impaired) phonological awareness (PA), resulting in reading and language difficulties. Alternative theories suggest that impaired PA may have more of a top-down effect on auditory processing. Methods: The current study examines RAP in children tested at two time points, average age 5.6 and average age 8.3, in order to test causal relations between RAP and PA in a path analysis. Additional hierarchical regressions examine how well RAP predicts reading ability when accounting for PA and vocabulary. Results: The path analysis indicates a top-down effect, such that PA has a larger impact on BM over time than the reverse. Regressions indicate no direct impact of RAP on reading ability. Discussion: The path analysis provides evidence against the auditory processing hypothesis and instead suggests that between the ages of 5 and 8 it is variability in early phonological representations that predicts subsequent lower-level rapid auditory processing.