Most clinical studies of companion animals are reasonably similar to human clinical studies in that the “experimental unit” is the individual study subject – animal or human. Aside from the obvious differences, such as verbal communication or even the number of legs, more subtle differences often arise, such as blood-pressure measurement, for which the usual method for cats or dogs has not yet been validated. When the type of primary endpoint, the study design, and correspondingly the statistical analysis methods in a companion animal clinical study have direct parallels with human clinical studies, the methods for estimating sample size per treatment group are essentially the same. The focus of the presentation will be on two multi-center clinical studies – one completed and one currently in progress wherein treatments are randomized, by necessity, to ‘clusters’ of animals, where the ‘cluster’ sizes are unknown in advance of the study.