The first year of the COVID-19 pandemic in Brazil provided one of the most severe examples of its impacts on health and society. The country had death rates above the global average and acute impacts in increased unemployment, poverty, and threats to food security marked along ethnic and social lines. This study asks how different degrees of vulnerability between Brazilian cities lead to varying survival probabilities of their population in the phases of the pandemic in the country. To answer this question, this research presents a descriptive and analytic exploration of the relationship between vulnerability and COVID-19 from February 2020 to February 2021. We describe this period in seven distinct phases, characterised by geographic units, vectors of virus transmission, and infected cases and fatality numbers. In this context, we implement an exploratory survival analysis of COVID-19 fatalities using the Kaplan-Meier estimator (KME) in a set of cities with different social vulnerability degrees. The KME is a common analytic tool in medicine, and we implement it in a geographic investigation to focus on the temporal dimension of the crisis and examine socio-territorial vulnerability. Our results present a clear association between vulnerability and COVID-19 deaths. Highly vulnerable cities show low survival probabilities, and there are statistically significant differences in survival probability between low- and high-vulnerability cities. Further research should advance by investigating spatio-temporal dynamics, providing fine-resolution empirical information, and addressing behavioural components related to COVID-19 cases and deaths in the Global South.