Surface temperature drives many ecological processes and infrared thermography is widely used by ecologists to measure the thermal heterogeneity of different species' habitats. However, the potential bias in temperature readings caused by distance between the surface to be measured and the camera is still poorly acknowledged. We examined the effect of distance from 0.3 to 80 m on a variety of thermal metrics (mean temperature, standard deviation, patch richness and aggregation) under various weather conditions and for different structural complexity of the studied surface types (various surfaces with vegetation). We found that distance is a key modifier of the temperature measured by a thermal infrared camera. A non-linear relationship between distance and mean temperature, standard deviation and patch richness led to a rapid under-estimation of the thermal metrics within the first 20 m and then only a slight decrease between 20 and 80 m from the object. Solar radiation also enhanced the bias with increasing distance. Therefore, surface temperatures were under-estimated as distance increased and thermal mosaics were homogenized at long distances with a much stronger bias in the warmer than the colder parts of the distributions. The under-estimation of thermal metrics due to distance was explained by atmospheric composition and the pixel size effect. The structural complexity of the surface had little effect on the surface temperature bias. Finally, we provide general guidelines for ecologists to minimize inaccuracies caused by distance from the studied surface in thermography.