The effect of road traffic noise in urban environments is an issue of social and scientific interest, due to its public health and economic impacts. Scientific literature showed a decrease in the level of tyre/road noise generated as temperature increases, but usually under standardised traffic conditions in non-urban environments. Based on a wide network for the hourly monitoring of road traffic flow, air temperature and noise levels across the city of Madrid (Spain), this work proposes and applies a new experimental methodology for studying the dependence of urban road traffic noise on temperature. This study was conducted under real-world traffic conditions involving a wide variability in urban configurations and in the type and state of preservation of vehicles, tires and pavements. From the analysis of data for a whole year, a time interval was identified (from Tuesday to Thursday and between 8 a.m. and 8 p.m.) in which the variability in road traffic flow for the whole city of Madrid was stable enough to allow for a linear regression study between temperature and noise levels from urban road traffic. The relationships found were highly significant (p ≤ 0.001) for data from all the noise monitoring stations, with values of higher than 20% and up to 42% for the explanation of the variability in the measured noise levels by temperature at most of the measurement points. The values of the slope coefficients at the noise monitoring stations ranged from −0.036 to −0.125 dB/°C, with an average value of −0.090 ± 0.011 dB/°C. These results are within the range of values reported in the scientific literature for experimental tests conducted under conditions of controlled or free-flowing traffic.