Stocked fry are the primary source of fish for the restoration effort for Connecticut River Atlantic salmon Salmo salar, yet it is unknown whether there is a developmental stage at stocking that yields optimal growth and survival or whether good growth and survival can be achieved across a wide range of fry developmental stages. To evaluate the effects of developmental stage on growth and survival, we stocked otolith-marked (thermal-banding patterns) fry from four (1996) or three (1997) developmental stages (thermally delayed, nominal, thermally accelerated-fed, and thermally accelerated-unfed) into three rivers in the spring and sampled the age-0 fish in the fall. There was no difference in 1996 among delayed, nominal, or accelerated-fed developmental stage treatments in the final size or population estimate. Few fish from the accelerated-unfed treatment were recovered. Results from a laboratory starvation study suggested that fish from the accelerated-unfed treatment would die from starvation in about 4 d after release. In 1997, accelerated-fed fish were recaptured at a slightly higher rate than were fish from the delayed or nominal treatment, but final sizes did not differ among treatments. Despite differences in population estimates and sizes among rivers, there was no interaction between treatment and river in either study year, indicating that treatment effects were consistent among rivers. Our results suggest that fry from a fairly wide range of developmental stages will survive equally well and will grow to similar sizes by fall unless the fish have been accelerated without feeding.