Abstract: Thermal-unit accumulation is commonly used to simulate crop phenology, because the crop growth rate depends mainly on the temperature in farmland. However, there is a great difference of thermal units that are derived from hourly and daily temperature, due to the diurnal variation of temperature. Therefore, this study aims to compare the simulation effects of two thermal units on crop phenology. The phenological data of summer maize and hourly temperature at four sites were collected from Zhengzhou, Nanyang, Huojia, and Huangfanqu Farm. The field experimental data in Zhengzhou ranged from 2005 to 2018, while the data at other sites was accessible for a period from 2012 to 2013. Three models of crop phenological rate in response to temperature were selected to simulate summer maize phenology, including linear, logistic, and Wang-Engel (WE) model. Subsequently, three cardinal temperatures of summer maize (the base, optimum, and the maximum temperature), the accumulations of the Hourly Thermal Units (HTU) , and Daily Thermal Units (DTU) were calculated in different phenological stages. The effects of two thermal units on summer maize phenology were compared for different models and phenological stages, including emergency, jointing, flowering, and maturity stage. Specifically, the model performance was evaluated using statistical indicators, such as variable coefficient, the difference between maximum and minimum (Rg), absolute root mean squared error (RMSE), normalized root mean squared error (NRMSE), and absolute bias (ABS) between simulated and measured values. The statistical indicators in phenological stages were also compared in the daily and hourly thermal units. The results showed that the DTU of the three models were all greater than HTU during the growing stage of summer maize, due directly to the diurnal variation of temperature. The maximum daily difference between DTU and HTU reached 9.7℃•d (Linear model), 9.1℃• d (Logistic model), and 7.4℃•d (WE model), respectively, when the daily average temperature was close to the optimum temperature for crop growth. Moreover, the correlation between HTU and DTU was the strongest in WE model (R2 = 0.927), followed by the logistic model (R 2 = 0.816), and the linear model (R2 = 0.738). The mean variable coefficient of HTU accumulation was 0.4%, smaller than those of DTU accumulation over the whole phenological period, indicating that HTU had higher stability than DTU. Furthermore, the DTU accumulation in the linear model was significantly greater (P<0.05) than HTU accumulation at jointing and flowing stages, while the DTU accumulation in the Logistic model was also greater (P<0.05) than HTU accumulation at jointing, flowing, and maturity stages. Nevertheless, there was no significant difference between DTU and HTU accumulation at each phenological stage in the WE model. The simulation of both DTU and HTU showed higher accuracy in the WE model than that in the Logistic model, followed by the linear model at phenological stages and intervals. The accuracies of three temperature models varied in the crop phenology with the root mean square error of 3.7 d, 3.9 d, and 5.1 d, and the NRMSE of 1.66%, 1.77% and 2.50% in the WE, Logistic and Linear models, respectively. In the term of accuracy differences at phenological intervals, the RMSE was 3.1, 3.3, and 3.9 d, and the normalized the root mean square error was 14.34%, 14.66%, and 17.74% in the WE, Logistic and Linear models, respectively. With the same temperature model, the differences between DTU and HTU accumulation were no more than 1d at a phenological stage, and 2 d in the phenological interval. The data demonstrated that there was little difference in thermal unit accumulation derived from hourly temperature and daily temperature for summer maize. Namely, there was no significant improvement in simulation accuracy of phenological stages with shorter time steps in HTU.