Lutzomyia longipalpis, the main sandfly vector for New World visceral leishmaniasis is a complex of an as yet undefined number of sibling species. At present, there is no consensus on the status (single species vs. species complex) of Brazilian populations. We applied five microsatellite loci to test the hypothesis that L. longipalpis occurs as two sympatric cryptic species in Sobral, Ceará State, Brazil as predicted by male sex pheromone chemotypes described previously for field specimens from this site [S-9-methyl-germacrene-B (9MGB) and a cembrene compound]. Abdominal spot morphology corresponds with pheromone type at this locality (9MGB in ‘1 spot’ males and cembrene in ‘2 spot’ males). Genotype data from 190 wild-caught L. longipalpis specimens collected in October 1999 and April 2001 were used to estimate genetic differentiation between the two sex pheromone populations and sampling dates. No significant (P > 0.05) genetic differences were found between the 1999 and 2001 9MGB samples (θ = 0.018; RST = −0.005), and genetic differentiation was low between the cembrene collections (θ = 0.037, P < 0.05; RST = −0.043, P > 0.05). By contrast, highly divergent allelic frequencies (largely at two microsatellite loci) corresponded to significant (P > 0.05) genetic differentiation (θ = 0.221; RST = 0.215) for all comparisons between samples with different pheromones. When pheromone samples were pooled across sample date, genetic differentiation was high (θ = 0.229; P < 0.001; Nem = 0.84). The allele frequency distribution at each of the five microsatellite loci was similar for males and females from the two collection years. Two of these loci showed highly divergent allele frequencies in the two sex pheromone populations. This was reflected in the highly significant genetic differentiation obtained from the male genotypes, between populations producing different pheromones (θ = 0.229–0.268; P < 0.0001 for the 2001 and θ = 0.254–0.558; P < 0.0001 for the 1999 collections, respectively). Similar results were obtained when the females, assigned to a pheromone type, were included in the analysis. Both a Bayesian analysis of the data set and a population assignment test provided strong evidence for two distinct populations corresponding to pheromone type. Given its genotype, the probability of assigning a 9MGB male to the original 9MGB population was 100% once the two years’ collections were pooled. For cembrene-producing ‘2 spot’ males this probability although still high, was lower than for 9MGB males, at 86%. This microsatellite data together with previously reported reproductive isolation between the two Sobral populations confirm that premating barriers are important in speciation of L. longipalpis.