The number of rice mills in Uganda increased rapidly during the past decade, in response to increasing demand for milling services. However, despite the notable improvement in access to milling services, recent studies have shown that some farmers still sell rice in unmilled form which attracts lower prices. This study was undertaken to examine why some rice farmers still sell un-milled rice in the advent of improved access to milling services. Descriptive statistical methods of data analysis were used to characterize rice-growing households by the form in which they sell rice, before fitting a Tobit model to determine the factors influencing the proportion of rice sold as grain after milling. The returns (gross margins) to rice-milling were also estimated. The study findings show that rice production is profitable regardless of the form in which it is sold; and the majority of households invest in milling all or part of their rice before sale. However, although milling households incurred higher costs, they also had higher gross margins, implying that selling milled rice is more profitable than selling paddy. The price of milled rice, volume of harvested rice, household size and group membership have significant and positive relationships with the proportion of rice sold as grain; while distance to the nearest rice mill is negatively and significantly associated with the proportion of rice sold as grain.