Several indices, including the normalized difference vegetation index (NDVI), normalized difference build-up index (NDBI), and normalized difference water index (NDWI), were retrieved from the Landsat 5 Thematic Mapper (TM) images, and the land surface temperature (LST) was reversed by the thermal radiation transfer model. Next, with the help of the thermal field variance index from LST, characteristics of spatial distribution of urban heat island were analyzed. Then, the regression models between LST and the three indices were formed by the least-squares method after they were aggregated from 30 to 120, 240, 480, 960, and 1200 m, and the relationships between the accuracies of the regression models and spatial scales were analyzed quantitatively. Finally, results from the experiment exemplified by Chongqing show that at all these spatial scales, both NDVI and NDWI are negatively correlated with LST and NDBI is positively correlated with LST. At the same spatial scale, cooling effect by NDWI acting upon LST is superior to that by NDVI, while inferior to that by NDBI to enhance LST. The fitting determination coefficients () of the regression models of LST and NDVI, NDBI, and NDWI increase from 0.23 (30 m) to 0.57 (1200 m), from 0.34 (30 m) to 0.70 (1200 m), and from 0.31 (30 m) to 0.68 (1200 m), respectively. Further, the of all of the regression models and spatial scales have positive logarithmic function relations, while the standard deviation and spatial scale have negative logarithmic function relations correspondingly, and the of all of the logarithmic models are . The heat islands of Chongqing are roughly along northeast–southwest directions, while the heat islands of the urban core area, such as Yuzhong district, are not obvious due to the influence of the Yangtze River and the Jialing River.