Canopy carbon flux was continuously measured in 2003-2013 using the eddy covariance system in a winter wheat field in Yucheng station, Shandong Province, to explore the effect of extreme heat on winter wheat canopy carbon assimilation. T_he extreme heat (EH) threshold was determined as the 95th percentile of daily maximum temperature (T_(a_max)), which was 30.58 ℃ during grain-filling stage in 2003-2013 in the study area. T_wo typical couples of EH and non-EH days in 2004 and 2012 were selected and compared to determine the characteristics and mechanism of the effects of EH on net ecosystem productivity (NEP, daytime), which was used to denote winter wheat canopy carbon assimilation rate. T_he results showed that T_(a_max) in the EH days were 3.10 ℃ and 3.17 ℃ higher than those in the non-EH days respectively in 2004 and 2012. T_hen total amounts of daytime NEP in EH days decreased by 3.25 mg(CO_2)?m~(-2)?s~(-1) (with decreasing rate of 19.77%) and 6.04 mg(CO_2)?m~(-2)?s~(-1) (with decreasing rate of 19.17%), respectively, compared with that in non-EH days in 2004 and 2012. NEP had a significant quadratic curve correlation with PAR, but no significant when PAR was greater than 1 000 mumol?m~(-2)?s~(-1), at that time NEP accounted for over 52.31% of total daytime NEP for both EH and non-EH days. T_he difference in NEP between EH and non-EH days increased with increasing PAR, especially when PAR was greater than 1 000 mumol?m~(-2)?s~(-1). T_here were no significant correlations between relative humidity of atmosphere (RH) and NEP, whether during whole day or when PAR was more than 1 000 mumol?m~(-2)?s~(-1). In the four observed days, 0-20 cm soil water content (SWC) was about 80% of field capacity. T_his was the appropriate soil water content at grain-filling stage of winter wheat, which had no negative effect on NEP. For the whole days, NEP had no significant correlation with air temperature (T_a) in both EH and non-EH days. However, during the daytime with PAR > 1 000 mumol?m~(-2)?s~(-1), NEP had significant negative correlation with T_a. NEP decreased by 7.28%-9.53% in 2004 and by 6.94%-10.42% in 2012 with every 1 ℃ increase in T_a. In conclusion, NEP was largely restrained by T_a when T_a was above 30.58 ℃ at the grain-filling process of winter wheat in the North China Plain. While the contribution of T_a to daytime NEP inhibition was 59%-83%, total NEP decreased by 6.05%-6.37% with 1 ℃ increase in T_a for EH days.