Wright and McPhee (1925) suggested a method of estimating the inbreeding coefficient of an individual based on the probability that a pair of lineages traced randomly, one through the maternal line and one through the paternal line, both contain a common ancestor. (One-half of this probability is an unbiased estimate of the inbreeding coefficient). In their procedure, maternal and paternal lines are chosen in pairs, and comparisons are made only between the lines in a pair. A more efficient procedure is to compare every maternal line with every paternal line, a procedure used by Robertson and Mason (1954). In this paper we provide estimates of the sampling variance of the inbreeding coefficient as estimated by the multiple comparison method, and we examine the relative efficiency of this method and the Wright-McPhee procedure. Formulae are also provided for ascertaining the optimal sampling method for estimating the average inbreeding coefficient of a group or herd.