To identify the environmental changes responsible for the declines in abundance shown by many granivorous bird species, the demographic mechanism through which the changes have acted must be determined. Ring-recovery data were used to estimate the annual survival rates (since 1962) of six seed-eating bird species with contrasting population trends to identify whether variations in survival could have been the mechanism behind population change. The survival rates of Bullfinch Pyrrhula pyrrhula, Chaffinch Fringilla coelebs, Goldfinch Carduelis carduelis, Greenfinch C. choris, Linnet C. cannabina and House Sparrow Passer domesticus were estimated using models allowing age- and time-specificity in survival (reporting rates could be assumed to be constant). Three tests of the importance of variations in survival in determining population trend were conducted: (1) simple population models with constant productivity showed whether temporal changes in survival were sufficient alone to explain observed trends in abundance, (2) survival models incorporating changes in abundance as a covariate identified whether annual survival rates were associated with population changes, and (3) mean survival rates found in objectively identified periods of increase, decline and stability in each species' population trend were compared. These analyses suggested that environmental change has led to the observed population trends for Goldfinch and House Sparrow largely through effects on survival. Weaker relationships between variations in survival and population trend were found for Bullfinch, Chaffinch and Linnet, but other factors such as breeding success are likely to have been at least as important for these species, and also for Greenfinch. Checking analyses incorporating density-dependence did not alter these conclusions.