Populations of Sancassania mites were collected in Costa Rica from scarabaeid and passalid beetles and cultured. The populations proved to be reproductively incompatible due to postzygotic isolation and show 1.4% difference in domains 2 and 3 of the 28S nuclear rDNA gene, indicating the populations represent distinct species. Because the mites were virtually indistinguishable morphologically, 61 morphological characters of 50 females and 101 characters of 60 deutonymphs of the two species were analyzed. Traditional univariate morphometrics could not separate them. Multivariate analyses of variance (principal component and discriminant function) were used to interpret morphological differences between the two species in relation to factors that influence their morphology in a laboratory and field setting. Principal component analyses were done on size and shape as well as shape variables alone. The discriminant function analysis was done on a reduced subset of shape variables. In both cases, the shape analyses resulted in complete separation of the two species and the characters contributing strongly to the discrimination were used in formal description of the two species, Sancassania salasi sp. nov. and S. ochoai sp. nov. Although deutonymphs of S. salasi taken from field-collected beetles show a significantly smaller magnitude of size variation, they show significant deviation in shape compared with cultured deutonymphs of the same species, a potential problem for correlation of specimens of other Sancassania species from culture and nature. Characters that provide the strongest contribution to these intraspecific shape changes are, therefore, taxonomically unreliable.