Landraces (LRs) are important as a source of novel alleles for crop improvement, and their conservation is therefore necessary for food security. These genetic resources have suffered continuous erosion, especially in more accessible areas. We assess the loss of Nepali rice LRs and identify the factors influencing the probability of cultivating the most dominant LR via a logistic regression model. The majority of farmers cultivate LRs and modern varieties simultaneously. However, there has been a decrease in varietal diversity and a loss of some LRs over recent years, mainly because of their low yield, their sensitivity to diseases and pests, and their late maturity. The opportunity cost of maintaining the Satha landrace is higher on irrigated farms and for those farmers specializing in niche products. On the other hand, Satha is more likely to be cultivated by large-scale farmers and by those having a religious and/or a cultural attachment to this LR. Market-based incentives are less costly than publicly funded conservation programmes, and the superior taste of some LRs may allow them to be developed as niche products. However, to achieve this, public investment is needed to generate the necessary support infrastructure. In the more accessible Terai region, there is a particular need to introduce a flexible incentive mechanism to maintain LRs and to offset the negative effect of development intervention.