Suggestions that short-faced members of the Felidae tend to lack the second upper premolar (P2) imply a possible shift in scaling associated with the palate and maxillary tooth row in Lynx, which lacks P2, as compared to felids that retain it. This hypothesis is tested using a scaling model that relates the lengths of the palate, and the upper tooth row and its components, to post-palatal skull length in the small to moderately large felids Felis catus (domestic cat), L. canadensis (Canada lynx), F. pardalis (ocelot), and F. concolor (cougar). Scaling relationships of both palate and tooth row length to post-palatal skull length do not differ significantly from isometry in all four species. However, ocelots have a significantly shorter palate and tooth row than lynxes over their overlapping ranges of post-palatal skull length, suggesting that the absence of P2 is not correlated with the length of the face in these species. CI, P3 and P4 tend to be relatively longer in larger felids; none the less, ocelots have a relatively small P3 and lynxes have a proportionately large P4. Because both lynxes and ocelots have a relatively small gap between CI and P3, the absence of P2 is not correlated with available space within the tooth row in adults. However, lynxes also appear to have a relatively long dP3 that almost obliterates the diastema within the deciduous tooth row. The absence of P2 in Lynx may be an engineering artefact that is associated with a shift in proportions within the deciduous toothrow, resulting in inhibition of the development of P2 and dP2 early in ontogeny. Despite the variable occurrence and polymorphism associated with P2 in the Felidae, this character has systematic value within this clade and is a synapomorphy for Lynx.