Multilateral development banks are frequently accused of"defensive lending, "the practice of extending new loans purely in order to ensure that existing loans are repaid. This paper empirically examine this hypothesis using data on lending by and repayments to the International Development Association (IDA), which is the largest provider of concessional development loans to low-income countries. The authors argue that key institutional features of IDA both (i) potentially create incentives for defensive lending, and (ii) enable particularly sharp tests of the defensive lending hypothesis. The authors find that there is a surprisingly robust partial correlation between disbursements on new IDA loans and repayments on existing loans. However, a closer look at the evidence suggests that defensive lending is unlikely to be a major explanation for this partial correlation.