The developmental processes that contribute to variation of morphological traits are the subject of considerable interest when attempting to understand phenotypic evolution. It is well demonstrated that most characteristics of tooth pattern can be modified by tinkering conserved signal pathways involved in dental development. This effect can be evaluated by comparing developmental models with naturally occurring variation within explicit phylogenetic contexts. Here, we assess whether evolutionary changes in lower molar (M) ratios among platyrrhines were channelled by alterations in the balance of activators and inhibitors as predicted by the inhibitory cascade (IC) model (Kavanagh et al. in Nature 449:427–432, 2007). Ordinary linear regression adjusted to M2/M1 versus M3/M1 ratios of 38 species of platyrrhines indicated that the slope and intercept were significantly different from the IC model. Conversely, when the phylogeny was incorporated into the regression analyses (PGLS), variation in molar ratios did not differ from the developmental model. PGLS also showed that changes in molar proportions are not an allometric effect associated with body size. Discrepancies between phylogenetically corrected and non-corrected analyses are mainly due to the departure of Callitrichines from the predicted values. This subfamily displays agenesis of M3 with higher than expected M2/M1 ratios, indicating that M3 fails to develop even when the inhibition by M1 on the subsequent molars is not increased. Our results show that evolution in molar ratios is concordant with slight changes in the proportion of activators and inhibitors that regulate molar development; however, other processes are required to account for variation in the number of teeth.