Larval behaviors and physical forcing mechanisms contributing to the transport of Dun- geness crab Cancer magister megalopae into inlets have been determined for many areas along the coastlines of Washington, Oregon and California, but have never been examined in Alaska. In 2004 and 2005, we observed transport of C. magister megalopae into Glacier Bay through patterns of depth, spatial and temporal variation among 3 sites located at 8, 12 and 18 km from the Bay's mouth using light traps. At each site, 96.5 to 99.4% of megalopae were captured by traps positioned within 1 m of the surface, and few were captured in traps within 1 m of the bottom (10 m depth). Megalopae abundance decreased with increasing distance from the mouth of Glacier Bay, with the highest abun- dance captured in Bartlett Cove, followed by the South Beardslee Islands, and very few megalopae were captured in the North Beardslee Islands. Transport processes were correlated with proxies for lunar cycle, tidal amplitude, maximum sustained wind speed and interactions among these effects. Few megalopae were captured at any site during 3 ⁄4 and full moon periods. High megalopae abun- dances in Bartlett Cove were associated with new to 1 ⁄ moons and tidal amplitudes of 300 to 499 cm. Pulses of megalopae in the South Beardslee Islands occurred during new and 1 ⁄ moons with tides of 500 to 599 cm and moderate winds. Variation in megalopae supply to sites in lower Glacier Bay may be explained by differences in hydrodynamics, even among sites separated by less than 10 km.