The rapid change of land use and land cover in the Metropolitan area of Tehran has influenced the distribution pattern of land surface temperature (LST). In this study, a spatial autocorrelation analysis is adopted to process the spatial-temporal changes of LST and Normalized Difference Vegetation Index (NDVI) in Tehran during the period of 1987 to 2010. Global spatial autocorrelation analysis revealed that global Moran’s I of LST has increased while that of NDVI has decreased with time. While the newly hot clusters of LST were appeared in west and southwest of Tehran, the spatial extents of the old hot clusters have been increased. The Local Indicators of Spatial Association (LISA) analysis confirmed the intensifying and expansion of the hot clusters and weakening of the cold clusters of LST. The spatial pattern of LST and UHI expanded toward west and southwest of Tehran and UHI effect was intensified. Based on LISA analysis, the occurrence of increasing trend of HH cluster of LST is related to the decrease of HH cluster of NDVI in Tehran. Accompanying with decreasing of vegetation cover, green cover cooling effect also decreased, and consequently, UHI effect was intensified from 1987 to 2010. The results demonstrate the usefulness of spatial autocorrelation technique for analysis of spatial-temporal changes of UHI and green cover in urban areas. Ordinary Least Squares regression (OLS) and Geographically Weighted Regression (GWR) were used to investigate the relationships between LST and NDVI. The results indicate that the performance of the GWR model is significantly better than OLS.