When females mate with different males, competition for fertilizations occurs after insemination. Such sperm competition is usually summarized at the level of the population or species by the parameter, P2, defined as the proportion of offspring sired by the second male in double mating trials. However, considerable variation in P2 may occur within populations, and such variation limits the utility of population-wide or species P2 estimates as descriptors of sperm usage. To fully understand the causes and consequences of sperm competition requires estimates of not only mean P2, but also intra-specific variation in P2. Here we investigate within-population quantitative variation in P2 using a controlled mating experiment and microsatellite profiling of progeny in the multiply mating stalk-eyed fly, Teleopsis dalmanni. We genotyped 381 offspring from 22 dam-sire pair families at four microsatellite loci. The mean population-wide P2 value of 0.40 was not significantly different from that expected under random sperm mixing (i.e. P2 = 0.5). However, patterns of paternity were highly variable between individual families; almost half of families displayed extreme second male biases resulting in zero or complete paternity, whereas only about one third of families had P2 values of 0.5, the remainder had significant, but moderate, paternity skew. Our data suggest that all modes of ejaculate competition, from extreme sperm precedence to complete sperm mixing, occur in T. dalmanni. Thus the population mean P2 value does not reflect the high underlying variance in familial P2. We discuss some of the potential causes and consequences of post-copulatory sexual selection in this important model species.