The conventionally used equilibrium displacement model (EDM) provides a convenient way to estimate the distribution of welfare gains from crop productivity growth across different population groups. However, it ignores the high market margin. Little research has been done on the potential bias in estimated welfare gains when the market margin is omitted. This study assesses ex ante how the welfare effects of genetically modified (GM) cassava in Uganda will be distributed across large-scale producers, small-scale producers and non-producers, and how much bias is embedded in the conventional EDM that ignores the market margin. The results indicate that welfare gains from GM cassava will be enjoyed by all three groups but most by large-scale cassava producers, and that the bias that results from ignoring the market margin is relatively small when looking at the benefits for the entire country, but more serious for the population groups regarded separately.