Experiments were conducted from 1992 to 1996 at four small-scale hydroelectric plants on salmon rivers in the Southwest of France to relate downstream bypass efficiency to hydraulic conditions and to the behavior of salmon Salmo salar and sea trout S. trutta smolts in the intake canal. The maximum turbine discharge varied from 20 to 85 m3/s and the width of intakes varied from 11 m to 30 m depending on the plant. The surface bypasses were located laterally along the intake at one end of the trashrack. The mean bypass discharges vary from 0.4 m3/s to 4 m3/s, or an average of 2 to 8% of the turbine discharge. The efficiency of the devices was evaluated by the mark-recapture technique. Radio telemetry was used to monitor movement patterns of salmon and sea trout smolts in front of the intake and near the bypass entrances. Depending on the site, the mean efficiency was found to be between 17 and 80%. Behavior of fish in the vicinity of the trashrack and the bypass seemed to be largely influenced by the flow pattern. Poor hydraulic conditions (turbulence, strong acceleration, upwellings) and insufficient discharge were identified through direct and video observations as being responsible for many aborted passages at the bypass entrances.The results suggest that siting of surface bypass systems must take into account the flow patterns in both the trashrack area and in intake canal. It is suggested that surface bypasses associated with existing trashracks may be an acceptable mitigation technology at small-scale hydroelectric projects where it is not necessary to garantee a highly efficient downstream passage protection.