The poor in developing countries are the most exposed to natural catastrophes and microfinance organizations may potentially ease their economic recovery. Yet, no evidence on MFIs strategies after natural disasters exists. We aim to fill this gap with a database which merges bank records of loans, issued before and after the 2004 Tsunami by a Sri Lankan MFI recapitalized by Western donors, with detailed survey data on the corresponding borrowers. Evidence of effective post-calamity intervention is supported since the defaults in the post-Tsunami years (2004-2006) do not imply smaller loans in the period following the recovery (2007-2011) while Tsunami damages increase their size. Furthermore, a cross-subsidization mechanism is in place: clients with a long successful credit history (and also those not damaged by the calamity) pay higher interest rates. All these features helped damaged people to recover and repay both new and previous loans. However, we also document an abnormal and significant increase in default rates of non victims suggesting the existence of contagion and/or strategic default problems. For this reason we suggest reconversion of donor aid into financial support to compulsory microinsurance schemes for borrowers.