Two populations of homing sockeye salmon (Oncorhynchus nerka; Adams and Chilko) were intercepted in the marine approaches around the northern and southern ends of Vancouver Island (British Columbia, Canada) en route to a natal river. More than 500 salmon were nonlethally biopsied for blood plasma, gill filament tips, and gross somatic energy (GSE) and were released with either acoustic or radio transmitters. At the time of capture, GSE, body length, and circulating testosterone ([T]) differed between populations, differences that reflected known life-history variations. Within-population analyses showed that in Adams sockeye salmon, plasma glucose ([glu]), lactate ([lactate]), and ion concentrations were higher in the northern approach than in the southern approach, suggesting that the former was more stressful. GSE, [T], and gill Na(+),K(+)-ATPase activities also differed between the two locales, and each varied significantly with Julian date, suggesting seasonality. Despite these relative geographic differences, the timing of river entry and the ability to reach spawning areas were strongly correlated with energetic, reproductive, and osmoregulatory state. Salmon that delayed river entry and reached spawning areas had relatively high GSE and low [T] and gill ATPase. In contrast, salmon that entered the river directly but that ultimately failed to reach spawning areas had lower GSE and higher [T] and gill ATPase, and they also swam at significantly faster rates (failed fish approximately 20.0 km d(-1) vs. successful fish approximately 15.5 km d(-1)). Physiologically, salmon that did not enter the river at all but that presumably died in the marine environment exhibited high stress (plasma [glu] and [lactate]) and ionoregulatory measures (plasma [Na(+)], [Cl(-)], osmolality).