Urban heat islands (UHI) can lead to multiple adverse impacts, including increased air pollution, morbidity, and energy consumption. The association between UHI effects and land cover characteristics has been extensively studied but is insufficiently understood in inland cities due to their unique urban environments. This study sought to investigate the spatiotemporal variations of the thermal environment and their relationships with land cover composition and configuration in Xi’an, the largest city in northwestern China. Land cover maps were classified and land surface temperature (LST) was estimated using Landsat imagery in six time periods from 1995 to 2020. The variations of surface heat island were captured using multi-temporal LST data and a surface urban heat island intensity (SUHII) indicator. The relationship between land cover features and land surface temperature was analyzed through multi-resolution grids and correlation analysis. The results showed that mean SUHII in the study area increased from 0.683 °C in 1995 to 2.759 °C in 2020. The densities of impervious surfaces had a stronger impact on LST than green space, with Pearson’s correlation coefficient r ranging from 0.59 to 0.97. The correlation between normalized difference impervious surface index and LST was enhanced with the enlargement of the grid cell size. The correlations between normalized difference vegetation index and LST reached maxima and stabilized at grid cell sizes of 210 and 240 m. Increasing the total area and aggregation level of urban green space alleviated the negative impacts of UHI in the study area. Our results also highlight the necessity of multi-scale analysis for examining the relationships between landscape configuration metrics and LST. These findings improved our understanding of the spatiotemporal variation of the surface urban heat island effect and its relationship with land cover features in a major inland city of China.