This thesis investigates the patterns and causes of childhood mortality in turn of the twentieth century Ontario and Toronto, Canada. There is growing emphasis in the historical demography and geography literature on the trends and spatial variations, as well as the determinants, of childhood mortality (deaths of children under the age of five). However, our understanding of these issues, and in particular the patterns and determinants within specific places, is limited by restrictions on the accessibility of individual-level data. This research focuses on documenting and explaining patterns of childhood mortality at three scales: the city/region, the neighbourhood, and the individual household. Using a case study approach, aggregate infant mortality rates for cities in Ontario between 1881-1941 were documented at five year intervals. Contrary to expectations, city size, population density and household crowding did not explain the variation in urban mortality rates. Instead, the socio-cultural composition of the population explained the greatest share of the mortality variation at the inter-urban scale. Analyses of Toronto in 1901 at the individual household and neighbourhood scales utilise data collected from individual-level death records, which were manually linked to the manuscript Census of Canada and tax assessment rolls. Results of the household level bivariate analyses reveal that social class (assessed value of home), housing conditions (crowding) and socio-cultural affiliation (religion, nativity, ethnicity) were independently associated with childhood mortality. Multivariate analysis confirmed these findings, and pointed to the importance of religion and crowding in particular. At the neighbourhood scale, analysis of the social geography of Toronto indicated a greater level of residential segregation by ethnicity and religion than previously understood. Childhood mortality varied a good deal from one neighbourhood to another, but not in the expected or predicted ways. Mortality rates were equally high in suburban fringe areas and inner city slums, and these variations could not be explained by social class or housing conditions, rather cultural affiliation had the strongest influence. In this regard, cultural background, but not necessarily immigrant status, was the most important determinant of mortality patterns at all three scales of analysis.