We carry out a randomized controlled experiment in West Bengal India to test three separate performance pay treatments in the public health sector. Performance is judged on improvements in child malnutrition. First, we exogenously change wages of government employed child care workers through a basic level of absolute incentives. The second treatment introduces high absolute incentives. Finally, we also test for the impact of basic relative incentives on child health. All treatments include supplying mothers with recipe books. The main results suggest that high absolute incentives reduce severe malnutrition by about 6.3 percentage points over three months. There are no signi…cant e¤ects on health outcomes of basic absolute or basic relative incentives. Results are robust to controlling for prior trends, propensity score matching, and reversion-to-the-mean. This result is consistent with a reported increase in protein-rich diet at home in the high absolute treatment.