In 2004 and 2005, steelhead (anadromous rainbow trout Oncorhynchus mykiss) reared at Winthrop National Fish Hatchery (Winthrop, Washington) were tagged with passive integrated transponders and subjected to a volitional or forced release. Gill Na+, K+‐ATPase (NKA) activity, body size, condition factor (K), travel time, and apparent survival were compared between volitional migrants (VMs), forced‐release migrants (FRMs), and volitional‐release nonmigrants (VNMs). Gill NKA activity and body size were similar between VNM, VM, and FRM groups, but K varied by date and migrant group. However, by the end of the volitional‐release period, all three groups had similar K. The VNMs consistently demonstrated lower apparent survival to McNary Dam on the Columbia River than did the VM and FRM groups, which had three to eight times greater survival than did VNMs. Volitional migrants had higher apparent survival to McNary Dam than did FRMs in 2005; however, apparent survival was similar between the two groups in 2004. Apparent survival between McNary and John Day dams was similar among all three groups in 2004 and 2005. Median travel time to McNary Dam was not consistently different among groups and differed between years. By combining results from VNMs and VMs, we were able to compare apparent survival between volitional‐ and forced‐release strategies. In 2004, the forced‐release group had a significantly higher apparent survival to McNary Dam than did volitionally released fish, whereas there was no difference in apparent survival between the two release strategies in 2005. There was little evidence for a survival‐, size‐, or physiology‐related advantage of volitionally released fish over forced‐release fish. The consistently lower apparent survival of VNMs does support the use of volitional release to screen fish that will fail to migrate downstream due either to low survival or residualization in the natal stream.