This study combines a livelihoods approach with a regression approach to quantify the effectiveness of irrigation infrastructure investment on improving people's livelihood strategies. Using a unique dataset based on households in southern Sri Lanka, and a natural experimental setting, we estimate from a two-stage income regression model to show that irrigation access has a positive effect on income through livelihood choices. We also show through qualitative approaches that factors not linked to irrigation infrastructure may contribute to changes in livelihood portfolios. In addition, we highlight factors that result in certain households being unable to move out of poverty despite access to the improved irrigation infrastructure.