The interregional variation coefficient (VC) has been considered as an accurate measure of the dispersion of regional infant, neonatal, postneonatal and perinatal mortality rates. Thus, trends of annual VC have been analyzed, for each rate, from 1940 to 1986, to identify the evolution in time of the regional differences with respect to these mortality rates. None of the four mortality rates showed a decreasing trend in their respective VCs. This may indicate that interregional differences do not change along time. The postneonatal mortality rate has been shown to have the highest VCs during the study period (ranging from 23 to 40%), with an independent evolution with respect to neonatal mortality rate, probably because the factors that influence both rates are clearly different.