The data of observations on the temperature and humidity of air in St. Petersburg and its environs are analyzed. The probability density functions are calculated for the temperature and water vapor pressure differences (ΔT and Δe, respectively) in St. Petersburg and two stations located in its environs. Analysis of these functions for day and night times and two seasons of the year as well as a statistical correlation between Δe and ΔT leads to a conclusion that meteorological factors and primarily variations of the effective emissivity of the underlying surface caused by the difference Δe and of the evaporation rate are decisive for the formation of the urban temperature field (heat island).