Clouds dominate energy balance and hydrologic cycle, regulate weather and climate. Evaluating model simulated cloud characteristics is essential for identifying the biases for improvement of cloud treatments in global climate models. Here, we examine cloud diurnal variation (CDV) that characterizes the contrast between day- and night-time cloud radiative effects. The study is based on the use of an index (daytime vs. daily-mean cloud fraction) proposed by Chen and Wang (2016, GRL). Analyses of 20-CMIP5 models reveal that, while the CDVs over oceans are all reasonably simulated when compare with observations, their values over land (notably deserts and plateaus) are consistently smaller, indicating smaller daytime clouds. The biases in CDV land-ocean contrast is found to be the main reason for poor spatial corrections in clouds between observation and models. For the CMIP5 models, the CDV bias, while consistent with the shortwave cloud radiative effect of the individual models, also explains the inter-model differences. Discussion on the effects of CDV biases on model simulations will be presented.