Extreme climate, especially temperature, can severely reduce wheat yield. As global warming has already begun to increase mean temperature and the occurrence of extreme temperatures, it has become urgent to accelerate the 5–20 year process of breeding for new wheat varieties, to adapt to future climate. We analyzed the patterns of frost and heat events across the Australian wheatbelt based on 50 years of historical records (1960–2009) for 2864 weather stations. Flowering dates of three contrasting-maturity wheat varieties were simulated for a wide range of sowing dates in 22 locations for ‘current’ climate (1960–2009) and eight future scenarios (high and low CO2 emission, dry and wet precipitation scenarios, in 2030 and 2050). The results highlighted the substantial spatial variability of frost and heat events across the Australian wheatbelt in current and future climates. As both ‘last frost’ and ‘first heat’ events would occur earlier in the season, the ‘target’ sowing and flowering windows (defined as risk less than 10% for frost (<0 °C) and less than 30% for heat (>35 °C) around flowering) would be shifted earlier by up to 2 and 1 month(s), respectively, in 2050. A short-season variety would require a shift in target sowing window 2-fold greater than long- and medium-season varieties by 2050 (8 vs. 4 days on average across locations and scenarios, respectively), but would suffer a lesser decrease in the length of the vegetative period (4 vs. 7 days). Overall, warmer winters would shorten the wheat season by up to 6 weeks, especially during preflowering. This faster crop cycle is associated with a reduced time for resource acquisition, and potential yield loss. As far as favourable rain and modern equipment would allow, early sowing and longer season varieties (i.e. in current climate) would be the best strategies to adapt to future climates.