Well-known formulae for the calculation of the necessary number of animals in strain comparisons are presented. Their use is illustrated with an example with fish. Use of the formulae in the planning and design of strain comparisons is recommended. When the necessary number cannot be attained and the experiment is conducted with fewer animals than what the formulae indicate, or the animals originate from a low and thus most likely non-random sample of parents, no inferences should be drawn about any observed between strain differences.