Diverse cultivars of winter wheat (Triticum aestivum L.) were grown in the field in 1993/94 and 1994/95 at Reading UK in temperature gradient tunnels at normal atmospheric (c. 370) or elevated CO2 concentration (c. 700 μmol CO2 mol−1 air). In 1993/94, grain yield of cv. Avalon was insensitive to mean temperature (between 8·8 and 10·9°C), while elevated CO2 increased yield by 1·3 t ha−1 (12·6%). In all other cultivars, warming reduced grain yield and CO2 increased grain yield. In 1993/94, in cvs Galahad and Mercia the effects of CO2 and temperature on yield were additive. However, for cv. Hereward in both years and for cv. Soissons in 1994/95, there were negative interactions between the effects of CO2 and temperature on yield: the maximum benefit of doubling CO2 to grain yield, 4·5 and 2·7 t ha−1 (65 and 29%) respectively, occurred at cooler temperatures; there was no benefit from doubling CO2 (i.e. 0%) once the temperature had increased above the seasonal mean by 2·2–2·6°C in cv. Hereward and by 1·3°C in cv. Soissons. The beneficial effect of doubling CO2 on grain yield in cvs Galahad, Hereward, Mercia and Soissons was negated by an increase in mean seasonal temperature of only 0·7–2·0°C. Warming decreased root dry mass at anthesis in 1994/95 while it increased at elevated CO2 (49 and 186%, coolest and warmest regime, respectively). Carbon partitioned to roots declined progressively with warming, while at elevated CO2 there was an average of 56% increase in allocation to roots. The relative impacts of both CO2 and temperature were greater on root dry mass than on either grain yield or total above-ground biomass, while the effects on grain and biomass yield varied considerably between cultivars, suggesting that the impact of rising CO2 and temperature are likely to be dependent on cultivar.