My thesis analyzes the determinants of the differential impact of natural disasters in Latin America and the Caribbean identifying the political, social and economic structures that determine social outcomes of natural disasters. It studies two adaptation tools, one individual (social capital) and one institutional (public spending and humanitarian aid) that could reduce the social impact of natural disasters. At aggregate level (countries) I conduct a time-series cross-section (TSCS) analysis, between 1960 and 2010, to analyze the impact of social capital, international aid, and public spending on the number of deaths caused by natural disasters. At individual level, I use logistic regression models to predict the probability of becoming poor after the Haitian and Chilean earthquakes in 2010. Findings at aggregated level confirm that higher the levels of social capital, disaster relief aid and public spending lower the number of casualties due to a natural disaster. However, the efficacy of these tools depends on the institutional framework of the country. Countries with higher democratization levels public spending is more effective in reducing the death toll after natural disasters. On the contrary, in more autocratic governments humanitarian aid becomes more efficient in reducing the death rate. The individual level results show that the social participation and network ties are fundamental in reducing the impact of the earthquakes on the levels of poverty with some differences between the countries. Regarding institutional mechanisms, both in Haiti and Chile the previous levels of public spending and aid within societies do help in mitigating the impact of disasters, however, results demonstrate the funding distributed afterwards encounter several limitations.