In spring 1996 and 1997, we studied the prototype surface bypass and collector (SBC) at Lower Granite Dam on the Snake River in Washington. Our objectives were to determine the most efficient SBC configuration and to describe smolt movements and swimming behavior in the forebay. To do this, we used hydroacoustic and radiotelemetry techniques. The SBC was retrofitted onto the upstream face of the north half of the powerhouse to test the surface bypass method of diverting smolts from turbines. The SBC had three entrances, with mean velocities ranging from 0.37 to 1.92 m/s, and it discharged 113 m3/s through its outlet at Spill Bay 1, which was adjacent to the powerhouse. Different SBC configurations were created by altering the size and shape of entrances. During spring 1996 and 1997, river discharge was well above normal (123 and 154% of average, respectively). Powerhouse operations caused a strong downward component of flow upstream of the SBC. Many smolts (primarily steelhead and secondarily chinook salmon) were observed actively swimming upward in the water column. There were four times as many smolts diverted from turbines per unit volume of water with SBC flow than with spill flow, which indicated that the SBC may be an especially important bypass consideration in moderate- or low-flow years. The highest SBC efficiency (the proportion of total fish passing through the north half of the powerhouse by all routes that passed through the SBC) for any configuration tested was about 40%. Although no single SBC configuration stood out as the most efficient, the horizontal surface and maximum area configurations, or some combination of the two, are worth further investigation because they were moderately efficient.