Migratory connectivity refers to the extent to which individuals of a migratory population behave in unison, and has significant consequences for the ecology, evolution and conservation of migratory animals. We made a retrospective assessment of the migratory connectivity of River Simojoki and River Kymijoki populations of Atlantic salmon Salmo salar L. by using stable isotope analysis of archived scales to identify the final feeding areas used before ascending rivers for spawning. We also tested differences in migratory connectivity between wild and hatchery-reared salmon and compared Carlin-tag recoveries with salmon scale stable isotope analysis as methods for studying salmon migrations. Stable isotope (C-13, N-15) values from the last growth region of scales from salmon caught ascending their natal rivers were compared via discriminant analysis with those from scales of salmon caught in different Baltic Sea areas during 19892011. Most River Simojoki salmon had likely fed in the Baltic Proper (mean SD for ascending fish probability 0.59 0.32) with secondary likely feeding areas in the Bothnian Sea (0.21 0.26) and the Gulf of Finland (0.20 0.27). Most River Kymijoki salmon had likely fed in the Gulf of Finland (0.71 0.42) with the Baltic Proper (0.29 0.41) a secondary feeding area. The results did not indicate the Bothnian Sea to be an important feeding area. The two salmon populations showed weak migratory connectivity and rather fixed areal preference throughout the record irrespective of wild or stocked origin. Although the results from the scale stable isotope analyses were broadly consistent with previously reported Carlin-tag recoveries, we argue that the stable isotope approach offers several important advantages in the study of salmon migratory behaviour.