The LSTC dataset contains land surface temperature data in China (about 9.6 million square kilometers of land) during the period of 2003-2017, in monthly temporal and 5600 m spatial resolution. It combines MODIS daily data, monthly data and meteorological station data to reconstruct the true LST under cloud coverage, and then the data performance is further improved by establishing a regression analysis model. The accuracy analysis shows that the reconstruction result is closely correlated with the in-situ measurements, with an average RMSE is 1.39 °C, an average MAE of 1.30 ° C and an R2 of 0.97. The dataset can be used for the spatiotemporal evaluation of LST and will be useful for high temperature and drought studies and food security.