Rural finance has long been an important tool for poverty reduction and rural development by donors and governments, but the impacts have been controversial. Measuring impact is challenging due to identification problems caused by selection bias and governments' targeted interventions, while randomised trial data are scarce and limited to contexts where little to no rural finance exists. Using an author-collected dataset, we provide insights on a large-scale long-lasting subsidised rice credit programme in Myanmar, one of the poorest and, until recently, most economically isolated countries in Asia. Identification relies on a fuzzy regression discontinuity design, exploiting an arbitrary element to the credit provision rule which is based on rice landholding size. Although we find little evidence that rice yield or output is increased, we do see that the programme has some positive effects on total household income, suggesting a positive spillover effect on other farm income activities.