Conodont fossils are highly valuable for Paleozoic biostratigraphy and for interpreting evolutionary change, but identifying and describing conodont morphologies, and characterizing gradual shape variation remain challenging. We used geometric morphometrics (GM) to conduct the first landmark-based morphometric analysis of the biostratigraphically useful conodont genus Neognathodus. Our objective is to assess whether previously defined morphotype groups are reliably distinct from one another. As such, we reevaluate patterns of morphologic change in Neognathodus P1elements, perform maximum likelihood tests of evolutionary modes, and construct novel, GM-based biozonations through a Desmoinesian (Middle Pennsylvanian) section in the Illinois Basin. Our GM results record the entire spectrum of shape variability among Neognathodus morphotypes thus alleviating the problem of documenting and classifying gradual morphologic transitions between morphotypes. Statistically distinct GM groups support previously established classifications of N. bassleri, N. bothrops, and N. roundyi. Statistically indistinct pairs of GM groups do not support literature designations of N. medadultimus and N. medexultimus, and N. dilatus and N. metanodosus, and we synonymize each pair. Maximum likelihood tests of evolutionary modes provide the first statistical assessment of Neognathodus evolutionary models in the Desmoinesian. The most likely evolutionary models are an unbiased random walk or a general random walk. We name four distinct biozones through the Desmoinesian using GM results and these align with previous biozonation structure based on the Neognathodus Index (NI) illustrating that Neognathodus-based biostratigraphic correlations would not change between GM or NI methods. The structural similarity between both biozonations showcases that determining GM-based biozones is not redundant, as this comparison validates using landmark-based GM work to construct viable biozonations for subsequent stratigraphic correlations. Although this study is limited to the Illinois Basin, our quantitative methodology can be broadly applied to additional genera to test taxonomic designations, interpret statistically-robust evolutionary patterns, and construct valid biozones for this significant chordate group.