During April-July 2000, we radio-tagged and released juvenile Chinook salmon (Oncorhynchus tshawytscha) and steelhead (Oncorhynchus mykiss) to evaluate a prototype surface flow bypass at Bonneville Dam on the Columbia River. The mock bypass, called a prototype surface collector (PSC), had six vertical slot entrances that were each 6 m wide and 12 m deep. The PSC was retrofitted to the upstream face of Bonneville Dam's First Powerhouse. Our objectives were to: (1) assess species-specific differences in movement patterns and behaviour of fish within 6 m of the face of the PSC, (2) estimate the efficiency and effectiveness of the PSC and (3) evaluate factors affecting the performance of the PSC. We found that 60-72% of the fish, depending on species, detected within 6 m of the PSC entered it. Of the fish that passed the First Powerhouse at turbines 1-6, 79-83% entered the PSC. Diel period was a significant contributor to PSC performance for all species, and day of year was a significant contributor to PSC performance for subyearling Chinook salmon. The PSC was twice as effective (%fish/%flow) as the spillway, passing 2.5:1 steelhead and subyearling Chinook salmon and 2.4:1 yearling Chinook salmon per unit of water. If fully implemented, the PSC would increase the percentage of fish that pass the First Powerhouse through non-turbine routes from 65-77% (without the PSC) to 76-85% (with the PSC), depending on species. Published in 2008 by John Wiley & Sons, Ltd.