An analysis of temporal stratification was performed in order to improve the accuracy of the annual noise levels estimation by limiting noise sampling to certain periods of time where less variability of the noise data is observed. The analysis consisted in finding a stratification related to certain temporal aspects such as seasons, months, workdays, weekends and a combination of them. It was shown that there is low influence in seasons and months but a temporal stratification was found within the week, where week days are stratified in working and weekend days, being the former the ones that has less variability. For pedestrian or leisure activity related streets, the stratification seems not to be as evident as for the others. This stydy was performed using a large database of noise data from 37 different sampling points and continuous measurements of up to five years.