Using measures of physical risk from climate change, we develop a methodology to allocate currency pairs according to a country’s vulnerability and construct portfolios with decreasing vulnerability to physical risk. We show that non-G10 currencies are more vulnerable to physical risk, have become less vulnerable over time, and that the vulnerability measure is correlated with higher losses from natural disasters. Portfolios exposed to currencies with decreasing vulnerability have exhibited positive abnormal returns, with the abnormal return coming from currencies that have relatively high levels of vulnerability. These results exist in non-G10 currencies, while no relation is found within G10 currencies.