In migratory bird species with a breeding range extending from warm temperate to arctic regions, populations from the latter often winter south of the temperate populations (so-called leap-frog migration). We suggest that this is explained by a great adaptive value for birds breeding in temperate areas with a relatively long and irregular spring to remain during winter within the same climatic regime close to their breeding sites in order to respond directly to the development in spring of favourable breeding conditions. The advancement of arctic spring is more regular, and arctic birds may well migrate far away from their breeding sites and still return at optimal breeding time by use of their circannual clock. This is necessary, since once the arctic birds have migrated out of the polar climatic zone, they are prevented from detecting spring progress at their breeding grounds. Migratory patterns are particularly well documented in different species of waders and lend support to this explanation. The Dunlin Calidris alpina has been reported to differ from the expected pattern, birds from W. Europe migrating to NW Africa while many birds from arctic regions winter in Europe. We have examined available evidence and found that this conclusion is unwarranted the Dunlins wintering farthest to the south in NW Africa probably originate from various areas with a predictable spring (Greenland, Iceland, N. Scandinavian mountains and N. Eurasia), as would be expected from our conjecture. Icelandic breeding populations of different wader species differ widely with respect to the position of their winter quarters in relation to wintering areas of temperate and arctic populations. This suggests that the predictability of Icelandic spring differs considerably between species with different ecological requirements. The regularity and predictability of spring at different major stop-over sites may also be of importance for the segregation of winter quarters between populations from different breeding areas.