During July and August 1995-1997, we used radiotelemetry to estimate the migration rate of 405 juvenile fall chinook salmon Oncorhynchus tshawytscha (mean fork length, 138-144 mm) through Little Goose Reservoir. Migration rates decreased significantly as fish approached the dam. Median migration rates in 1995 were 26.0 km/d through the 45.9-km reach immediately below Lower Granite Dam, 14.9 km/d through the next 14.4 km, and 0.8 km/d in the Little Goose Dam forebay (0.6 km). Median migration rates through the same reaches were consistent among years: 24.8, 13.4, and 0.8 km/d in 1996 and 20.2, 10.2, and 1.0 km/d in 1997. Most fish migrated through the upper 45.9 km within 5 d and through the lower two reaches (15.0 km) within an additional 5 d. However, 10% to 20% of the fish spent a week or more in the forebay and lower reservoir. Radio-tagged smolts displayed two behaviors after entering the forebay: crossing the forebay and upstream excursions. Study fish crossed the forebay an average of 0.6-1.0 time/h, and 157 upstream excursions were identified, 15 of which were at least 14.4 km in length. Fish behavior in the forebay was associated with declining water velocities near the dam. Detections of passive integrated transponder tags suggest that similar delays occur in other lower Snake River reservoirs. Based on studies from the Columbia River, delays for 20% of the juvenile fall chinook salmon outmigrants in each of these forebays may have contributed to high predation losses and pose a serious challenge to efforts aimed at restoring this threatened salmon stock.