Studies on elevation gradients in Panama and Costa Rica have shown that leaf-litter ants exhibit a mid-elevation peak in diversity. This diversity pattern has been observed in other groups and regions, but uncertainty remains as to just how pervasive it is and what might explain it. Here we examine the robustness of the mid-elevation peak in ant diversity across the entire Middle American corridor, from Veracruz, Mexico, to Costa Rica. We sampled 56 sites distributed throughout Middle America. All were in closed-canopy evergreen wet forest, spanning 11° latitude, from near sea level to 2600 m elevation. Ants were extracted from 100 litter samples from each site and identified to genus or species. Model selection was performed on richness and diversity variables to test if ant diversity best fits a linear model or one allowing for a mid-elevation peak. Linear models were also used to examine the relationships among diversity measures and temperature, precipitation, and seasonality. Species richness measures and diversity indices that incorporate relative abundance show a similar relationship to elevation throughout the region: a truncated bell curve with a mode near 400 m. A cubic relationship is statistically favored over quadratic or linear. Temperature is a significant correlate with diversity, but does not predict a bell-curve. Precipitation and precipitation seasonality fail to explain much of the variability, and no combination of environmental variables predicts a bell curve. Potential causes of the truncated bell curve include lowland biotic attrition, mid-point attractors, and ecotonal transitions from lowland to montane communities. Analysis of 17 subclades within ants mostly showed the same truncated curve but six clades were anomalous. Distinctive behavioral or historical features potentially explain their patterns.