This case study is based on a survey of the productivity of Ankole cattle raised under pastoralist or agro-pastoralist conditions in an area of south-western Uganda. The aim of the survey was to estimate the average milk performance of these indigenous cattle and to compare it with that produced by other African indigenous populations in pastoral and agro-pastoral systems. A subsidiary objective was to use the results to identify a group of high yielding cows from the sampled herds that might be used for future breeding purposes. This is the type of approach that might theoretically be adopted to develop a nucleus breeding herd that could be used to promote genetic improvements in milk production by providing offsprings to farmers.The case study first describes how such a study can be designed and then demonstrates how a sample of the highest yielding cows can be selected taking into account the different ecological and production system environment of each herd.Least squares analyses of variance are undertaken to estimate herd effects; the data are then corrected for these effects by saving the residuals. By sorting these residuals in descending order the top animals can then be selected with the effect of herd removed.