Amphidromy, the dispersal of larvae to the estuarine or marine environment with juveniles recolonising fresh waters to complete development, is common on tropical islands. This has led to the suggestion that amphidromy is an adaptation to hydrologically unpredictable environments, allowing recolonisation after local extinction (dispersal-limitation hypothesis). Alternatively, amphidromy may be more common in streams with stronger flow, as high tractive forces flush larvae into the estuarine/marine environment, forcing them to recolonise (passive dispersal hypothesis).We tested these hypotheses in a continental gobiid fish, the Roman Nose Goby (Awaous acritosus) that inhabits hydrologically and physically diverse catchments on the east coast of Australia. We measured 87Sr/86Sr along transects of the otoliths of 92 adult fish from 12 catchments, identified migratory life histories from these data and correlated the proportion of migratory fish with key environmental variables. This information was augmented with data from mtDNA (n = 276) and six microsatellite loci (n = 429) from 19 catchments, to explore gene flow among catchments and to assess whether local adaptation was likely in the face of high or restricted gene flow.Estimates of 87Sr/86Sr along transects revealed high variation in life history, with 63% of individuals showing no evidence of marine/estuarine residence. The slope of catchments was the only significant variable associated with the proportion of larvae retained in catchments, with steep catchments having a greater proportion of adults exhibiting a marine/estuarine phase during early life. Total panmixia was detected with mtDNA and six microsatellite loci, suggesting that gene flow is high for this species.These results support the passive dispersal hypothesis. Steep slopes are likely to have higher tractive forces in their flow, leading to a greater proportion of larvae being flushed into the estuary or sea. In addition, high gene flow would inhibit local adaptation, making the dispersal-limitation hypothesis unlikely.This study provides a framework for understanding how local adaptation may be constrained by gene flow across the landscape so that the adaptive or passive mechanism underlying facultative amphidromy can be explored.