Aim Changes in climate and land use practices have been found to affect animal populations in different parts of the world. These studies have typically been conducted during the breeding season, whereas the non‐breeding season (hereafter ‘winter’) has received much less attention. Changes in regional winter abundances could be caused by changes in overall population sizes and/or redistribution of populations. We tested these mechanisms for terrestrial winter bird population changes in Northern Europe and explored the role of climate change and species habitat preference. Location The Netherlands, Denmark, Sweden, Finland. Methods We used winter bird counts from four countries conducted annually between 15 December and 20 January in 1980/1981–2013/2014. We report national population trends for 50 species for which a trend could be calculated in at least three of the countries. We analysed country‐specific population growth rates in relation to species’ climatic summer and winter niches, habitat preference and migratory behaviour. Results Species breeding in colder (typically northern) areas showed more negative winter population trends than species breeding in warmer areas. Regional winter population trends were negatively correlated with characteristics of their winter climatic niche: populations in the colder part of their winter distribution increased in abundance, whereas populations in the warmer part of their winter distribution decreased. Woodland species tended to do better than farmland species. Migratory behaviour did not explain variation in population trends. Main conclusions The generally decreasing winter population trends of cold‐dwelling breeding species probably reflect the general decline in population sizes of these species. In contrast, increasing winter population trends for populations in the colder parts of the winter distribution indicate a redistribution of wintering individuals towards the north‐east. Both these patterns are likely caused by climate change.

The relationships between species abundance or occurrence versus spatial variation in climate are commonly used in species distribution models to forecast future distributions. Under "space-for-time substitution", the effects of climate variation on species are assumed to be equivalent in both space and time. Two unresolved issues of

No abstract