In a 2003 study by Pacific Northwest National Laboratory (PNNL) for the U.S. Army Corps of Energy, we sampled nine gatewell slots at Bonneville Dam Second Powerhouse (B2) with a Dualfrequency Identification SONar (DIDSON) acoustic imaging device to estimate the gap loss of juvenile salmonids. Gap loss is the number of fish guided by screens but lost to turbine passage through the gaps between the tops of submerged traveling screens (STSs) and the intake ceilings. Six of the intakes (Units 15 and 17) had been modified to improve fish guidance efficiency (FGE, the proportion of fish passing above intake screens) while the three unmodified intakes at Unit 13 served as controls. All three units had similar configurations of turbine intake extensions (TIES). Intake modifications included removal of concrete between the gatewell and bulkhead slots to increase the area of the vertical barrier screen and installation of a turning vane and gap-closure device to direct more flow up into the gatewell slot.This study was to determine if those modifications, which did increase FGE, had the added benefit of reducing gap loss. In the unmodified intakes of Unit 13 we also sampled with infrared optical cameras to evaluate the proportions of fish and non-fish objects passing through the STS gaps and found that fish composed just 28.6% of all objects in spring and 12.9% in summer. Experiments in a laboratory tank confirmed that the DIDSON detects echoes from the surfaces of waterlogged sticks, macrophytes, and other debris as well as from fish. We developed filters based on target size, motion, range at first appearance, and the number of frames in which a target was seen to discriminate between fish and nonfish images. Filtered data produced estimates within 6% of those obtained by multiplying unfiltered DIDSON counts by the fish fraction estimated from optical-camera data.Results suggest that the intake modifications at Units 15 and 17 reduced gap loss relative to rates at unmodified Unit 13 by about 67% in spring and summer. An analysis of variance of differences in gap loss among units with modified and unmodified intakes (n = 9) indicated significantly higher gap loss in unmodified units than in modified units. In spring, the least-square mean rate for unmodified Unit 13 was 11.4% and this rate was significantly higher (P = 0.0001) than rates of 3.8% at modified Unit 15 and 3.6% at modified Unit 17, which did not differ significantly. In summer, the least-square mean rate for unmodified Unit 13 was 12.6% and this rate was significantly higher (P = 0.0188) than rates of 5.8% at modified Unit 15 and 3.4% at modified Unit 17.In spring, unmodified Intakes 13B and 13C had higher gap losses than all other intakes, and unmodified Intake 13A gap loss also was higher than were those of all modified intakes, except perhaps that of 15A, which lacked a gap-closure device. In summer, unmodified Intakes 13B and 13C also had higher gap losses than all other intakes, which did not appear to differ significantly from each other, except perhaps...