The use of Manly's (1985) statistic beta, in mate choice experiments is discussed. Beta is a measure of the homogamic mating preference of one sex (usually females) of a strain. Beta has several advantages over previously proposed measures of deviation from random mating: (a) its sampling distribution is known and hence it can be used in establishing confidence intervals and in hypothesis testing; (b) this distribution is continuous and unimodal, even when fairly small numbers of individuals are used; (c) beta is not affected by changes in the numbers of individuals used (provided the preference remains constant); (d) it is not affected by changes in the ratios of the two strains, and (e) it is not affected by the proportions of females mated. Several properties of beta and its estimate from an experiment, beta, are derived and have been verified for a wide range of conditions using computer simulations. Their consequences for experiments attempting to quantify sexual isolation are discussed. Finally, statistics to measure the significance of beta values are given.