Accurate hourly clear-sky irradiance (CSI) estimation is a crucial factor for most solar technologies in improving cost competitiveness and ensuring supply–demand balance. Numerous models have been developed to estimate CSI, including Clear-sky Irradiance models (CSIM) and Machine Learning (ML) models. In this study, 61 CSIM and 23 ML models for estimating hourly CSI are evaluated across the land of China, using hourly solar irradiance measurements observed at 35 stations of the China Ecological Research Network (CERN). The results reveal that the ML models generally obtain outperformed accuracy than other CSIM models. Among all selected models, the GPR3 model, Integretion3 model and Integretion2 model attain generally better ranks in China and each climatic zone, with global performance indicators (GPI) values of 31.66, 30.27 and 23.01, respectively. The GPR3 model is the top-ranked model in MPZ, SMZ, TCZ and TMZ, while the best model for TPMZ is the Integration3 model. In terms of the CSIM models, the Iqabal_C model is the most worth recommending model for hourly CSI estimation in China. The Kasten_I model is the top-ranked model in MPZ, SMZ and TCZ with GPI values of −1.23, 5.94 and −0.20. The Bird model and Iqbal_C model are the best models for hourly CSI estimation in TMZ and TPMZ with GPI values of 5.69 and 1.92, respectively. This study can offer guidance for the hourly CSI estimation models selection for different climatic zones in China.