The aim of this study was to analyse the association between season of birth, temperature and neonatal mortality according to socioeconomic status in northern Sweden from 1880 to 1950. The source material for this study comprised digitised parish records combined with local weather data. The association between temperature, seasonality, socioeconomic status and neonatal mortality was modelled using survival analysis. We can summarise our findings according to three time periods. During the first period (1880–1899), temperature and seasonality had the greatest association with high neonatal mortality, and the socioeconomic differences in vulnerability were small. The second period (1900–1929) was associated with a decline in seasonal and temperature-related vulnerabilities among all socioeconomic groups. For the last period (1930–1950), a new regime evolved with rapidly declining neonatal mortality rates involving class-specific temperature vulnerabilities, and there was a particular effect of high temperature among workers. We conclude that the effect of season of birth on neonatal mortality was declining for all socioeconomic groups (1880–1950), whereas weather vulnerability was pronounced either when the socioeconomic disparities in neonatal mortality were large (1880–1899) or during transformations from high to low neonatal rates in the course of industrialisation and urbanisation.