Mexican prehistory is characterized by migration and population isolation in its earliest phase, followed by later inter‐population cultural interactions, such as politics and trade. While shared, common morphological variants are often informative about recent population events, rare trait variants have not been widely investigated to see whether they might be informative about earlier events. Here, we consider populations that show several rare variants at frequencies high enough to warrant such an investigation. Examining past population structure can help us understand population interactions across different periods of time, levels of migration, and population isolation/drift. Multiple‐origin populations may have more variation, including more and higher rates of rare traits. We examined maxillary lateral incisors in 1117 dentitions from 76 samples of Preclassic/Classic and Postclassic pre‐European contact Mexico and the American Southwest for the prevalence of rare lateral incisor variants. Variants observed included barrel, congenital absence, interruption groove, mesial marginal bending, peg, reduced, supernumerary, and talon teeth. The most common variant overall is interruption groove. Central Mexico, Huasteca, and the Lowland Yucatan region samples present the highest overall number of incisor variants. Higher frequencies of single unusual traits are commonly found in samples from smaller populations, while samples from larger population centres show a greater range of these rare variants. We investigated to see whether the pattern of rare incisor variant frequencies reflected early or later population history by comparing similarity/distance matrices and factor model matrices using Mantel tests and Generalized Procrustes analysis. Results show a weak relationship with the Postclassic period and shared migration/language and trade interaction models. We suggest that later cultural interactions have acted to mask earlier population history. Research on serial founder effects should be approached multiregionally and across time, to avoid missing inter‐region biological relationships. Copyright © 2015 John Wiley & Sons, Ltd.