The metropolitan region of Barcelona (BMR) is one of the most densely populated areas in the Mediterranean countries. The estimation of air temperature at a short scale from satellite measurements would contribute to a better understanding of the varied and complex spatial distribution of temperatures in BMR. This estimation would be a first step to study several patterns of the thermometric regime affecting population life quality and health. Taking advantage of MODIS data, air temperature measurements at 48 thermometric stations along the year 2015, together with their geographic and topographic data, multiple regression analyses have permitted to obtain fine spatial distributions (pixels of 1 km2) of minimum, mean and maximum daily air temperatures. Previous to the multiple regression, Pearson coefficients and principal component analysis offer a first overview of the relevance of the variables on the empiric temperatures. The most relevant variables on the multiple regression process at annual and seasonal scale are land surface temperatures, latitude, longitude and calendar day. At a monthly scale, altitude (maximum temperature) and continentality (cold months for minimum and hot months for maximum temperatures) are also relevant. The best fits between empiric temperatures and those derived from the multiple regression processes have square regression coefficients within the range (0.92–0.96) for the annual case, (0.70–0.92) at seasonal scale and (0.52–0.87) at monthly scale. The root mean square error varies from 1.5 to 2.0 °C (annual case), from 1.3 to 2.0 °C (seasonal scale) and from 1.2 to 2.1 °C (monthly scale). In agreement with these regression coefficients and mean square errors, the obtained spatial distribution of temperatures is of notable quality. As an outstanding application, the detection of several urban heat islands on different conurbations within BMR along the Mediterranean coast becomes possible.