Annual variability and trends in winter wheat yields were evaluated for relationships toclimate in 6 long-term experiments (in total 1915–2005) and regionally in 3 counties (1965–2008) insouthern and central Sweden. The annual yield was predicted as function of a climate index and year.Alternative indices were used based on either monthly temperatures or yield predicted by a simpleweather-driven crop growth simulation model. The main results were as follows. (1) The yield predictions were better for regions than for the long-term experiments. (2) The time variable accounted formore of the yield trend in regions than in the experiments, and more in southern than in centralSweden. (3) The models based on a simple temperature index were often better yield predictors thanthe growth model. (4) The relations to winter temperatures became gradually weaker after ca. 1970.(5) The yield relation to simple temperature indices differed between locations, similarly for experimental and regional yields. (6) The strongest influence of climate was estimated by means of thegrowth model predictions of regional yields in Gotland County. (7) The fraction of regional yieldtrends (1965–1996; ranging from 57 to 90 kg ha–1 yr–1) that could be attributed to climate was estimated greatest in the Gotland region (27 to 64%). For climate change assessments of winter wheatyield in central Sweden, the growth model considering temperature, radiation and water conditionsduring spring and summer would also need to include processes of overwintering.