Sex determination in haplo-diploid animals has been explained by P. W. Whiting's (1939) single-locus, multiple-allele scheme, which is applicable to two cases only, or by da Cunha and Kerr's (1957) genic-balance scheme, a more general hypothesis. I propose a general hypothesis of sex determination in haplo-diploidy, based on Snell's (1935) multiple-factor suggestion. This multiple-locus hypothesis is that, in haplo-diploid species, sex is determined by a number of loci. Females are heterozygous at one or more loci, while males are homozygous or hemizygous at all sex loci. At the molecular level, this effect might be due to female-determining properties of heteropolymers formed betwen the products of different alleles at any sex locus; homopolymers or heteropolymers between products of different loci are not formed or lack sex-determining activity. Haploid intersexes could arise from mutants that form active homopolymers or active heteropolymers with products of other loci. Diploid intersexes should be extremely rare, except in single-locus species, in which intersexes could result from mutations that reduce heteropolymer formation. The data from a number of cases generally support the multiple-locus hypothesis for Hymenoptera and haplo-diploid mites but not for coccids. No suitable data exist for other haplo-diploid groups. Compared with single-locus species, those with many sex loci will have weaker selection operating on the alleles at each locus and will lose fewer diploids as low-viability males. Testable predictions for species with many sex loci indicate that prolonged close inbreeding should yield diploid males; that diploid intersexes in outbred lines should be extremely rare compared with haploid intersexes; and that feminized borders, due to complementation between different sex alleles, should often occur between genetically different blocks of tissue in gynoid males.