Salmon monitoring programs often measure juvenile production by operating migrant traps downstream of spawning and rearing areas during smolt migration. However, this approach does not account for individuals that move downstream of trapping locations prior to smolt sampling. We used a mark–recapture study with passive integrated transponder tagging to estimate the proportion of Coho Salmon Oncorhynchus kisutch juveniles, tagged in the fall in a Northern California stream, that migrated to rearing habitat downstream of a seasonally operated trap before spring smolt sampling. Emigrants were detected by using the migrant trap, located near the upstream limit of tidal influence, and continuously operated antennas located in tidal wetlands downstream of the trap. For all three cohorts sampled (2010, 2011, 2012), we identified two distinct emigration periods (not including fry emigrants that emigrated in spring at a size too small to tag): a fall–winter period, when early emigrant parr moved into a restored tidal wetland (early emigrants); and a spring period, when smolts emigrated (smolt emigrants). There was little movement in the intervening period. Emigration timing varied depending on the location in the basin where fish were tagged; locations in the lower main stem generally produced more early emigrants, while locations in the upper basin produced more smolt emigrants. Across locations, early emigrants accounted for 2–25% of the fall‐marked juveniles from 2010, 8–29% from 2011, and 7–13% in 2012. Smolt emigrants accounted for 15–49% of the fall‐marked juveniles from 2010, 13–14% from 2011, and 3–35% from 2012. The consistent occurrence of early emigration in this and other recent studies brings into question estimates of smolt abundance and demographic rates (e.g., overwinter and marine survival) that do not account for this life history variant. Received June 19, 2014; accepted October 23, 2014