SUMMARYThe degree of linkage disequilibrium, D, between two loci can be estimated by maximum likelihood from the frequency of diploid genotypes in a sample from a random-mating population. Haploid genotypes can be identified directly in some species from a sample of chromosomes extracted from the population and made homozygous, or by test crossing. The maximum likelihood estimators of D are described, with examples, for both methods, including the cases where both loci are codominant and one or both are dominant.The efficiencies of the methods are compared when D = 0; If both loci are codominant the estimate of D has the same variance,from a sample of .N identified diploids as from N identified haploid types, where p and q are the gene frequencies; therefore the diploid method is more efficient in practice since less labour is required. With dominance at either locus V(O) is lower for samples of the same size using the haploid method if the dominant alleles are at high frequency.