The relationship between initial population density of Globodera pallida and yield was examined for five genotypes of potato at four different experimental sites; two in Scotland and two in England. The experiments at each site were conducted in two stages. The experimental area was first manipulated using resistant and susceptible potato cultivars to give plots with different nematode population densities to act as the initial populations (P(I)) for the second stage. The main experiment was conducted using a range of host genotypes that differed in their tolerance and susceptibility. The utility of an inverse linear model relating yield to initial population density was confirmed as were differences in cultivar tolerance. Site differences in yield loss were also observed. It was found that the model could be modified to include the partitioning of genotype and site effects. The implications of this in terms of developing further models to predict yield loss and nematode population dynamics are discussed.