We used recently developed microsatellites to directly estimate inbreeding levels in two pairs of coexisting cryptic fig wasp species ('Pegoscapus hoffmeyeri sp. A and sp. B', 'P. gemellus sp. A and sp. B'). Previous tests of Hamilton's local mate competition (LMC) theory in fig wasps have used the number of dead foundresses in a fig fruit to indirectly estimate the relative contribution of each to the common brood and thereby the level of local mate competition. Further, the population level of inbreeding has been indirectly estimated using the distribution of foundress numbers across broods. Our direct genetic estimates confirmed previous assumptions that the species characterized by lower foundress numbers showed higher relative levels of inbreeding. However, there were quantitative differences between the observed level of inbreeding and the expectation based on the distribution of foundress numbers in both pollinator species associated with Ficus obtusifolia. Here, genotype compositions of broods revealed that only 23% of fruits with multiple foundresses actually contained brood from more than one foundress, thus explaining at least part of the underestimate of actual sibmating. Within the four wasp species there was no evidence for genetic differentiation among the wasp populations sampled from different trees across 20 km and from different points in time. Further, no genotypic disequilibrium was detected within any of the species. Although F1 hybrids were observed between the two species pollinating F. obtusifolia, there was no evidence of genetic introgression. Finally, we found that 11% of the sons of allospecifically mated mothers were diploid hybrids suggesting a break down of the sex determination system in hybrids.