We use three allopatric populations of the stalk-eyed fly Teleopsis dalmanni from Southeast Asia to test two predictions made by the sex chromosome drive hypothesis for Haldane's rule. The first is that modifiers that suppress or enhance drive should evolve rapidly and independently in isolated populations. The second is that drive loci or modifiers should also cause sterility in hybrid males. We tested these predictions by assaying the fertility of 2066 males derived from backcross experiments involving two pairs of populations and found that the proportion of mated males that fail to produce any offspring ranged from 38 to 60% among crosses with some males producing strongly female-biased or male-biased sex ratios. After genotyping each male at 25-28 genetic markers we found quantitative trait loci (QTL) that jointly influence male sterility, sperm length, and biased progeny sex ratios in each pair of populations, but almost no shared QTL between population crosses. We also discovered that the extant X SR chromosome has no effect on sex ratio or sterility in these backcross males. Whether shared QTL are caused by linkage or pleiotropy requires additional study. Nevertheless, these results indicate the presence of a "cryptic" drive system that is currently masked by suppressing elements that are associated with sterility and sperm length within but not between populations and, therefore, must have evolved since the populations became isolated, i.e., in ,100,000 years. We discuss how genes that influence sperm length may contribute to hybrid sterility.A major challenge to those who study speciation is to determine the causes of early reproductive isolation between incipient species. One avenue for gaining insight into this issue is to investigate the causes of Haldane's rule, i.e., the observation that the heterogametic sex of hybrid offspring are more likely to be sterile or inviable than the homogametic sex (Haldane 1922). This phenomenon has been observed in a wide range of animal taxa and is believed to be a nearly ubiquitous phase of early speciation (Orr 1997). Given that alleles for sterility or inviability are expected to be selected against within populations, models for the evolution of hybrid dysfunction typically assume two or more genes, each of which is neutral or advantageous within a population but have deleterious joint effects when mismatched between populations. Two nonmutually exclusive explanations (Turelli 1998) have been proposed for why these epistatic interactions, known as Dobzhansky-Muller incompatibilities (Dobzhansky 1937;Muller 1940Muller , 1942Orr 1995), should arise more quickly in the heterogametic sex.The dominance hypothesis (Muller 1942;Orr 1993;Turelli and Orr 1995) assumes that if genes causing hybrid dysfunction are recessive, then the heterogametic sex will be affected more than the homogametic sex. This result is expected because the degenerate (or missing, in XO taxa) sex chromosome possessed by the heterogametic sex fails to mask recessive alleles present on it...