Low-temperature thermochronologic data are useful for estimating mountain erosion rates because late stage cooling in mountain ranges often reflects exhumation via geomorphic erosion processes. Traditional interpretations of thermochronologic data assume steady, uniform erosion, conditions often violated in real mountain ranges. Here I examine whether unsteady or nonuniform erosion histories are detectable in patterns of erosion rate estimates made by traditional means. To do so, I numerically model the subsurface temperature field in a two-dimensional section of Earth's crust under various illustrative geomorphic scenarios. I generate simulated cooling ages within two isotopic systems, apatite fission track (FT) and (U-Th)/He, (He) at two landscape locations, ridge tops and valley bottoms. From these ages I make field estimates of erosion rates using the commonly employed elevation dependence and mineral pair methods. Calculations show that if errors on erosion rate estimates are kept small by replicate or transect sampling, many unsteady or nonuniform erosional histories do produce distinctive patterns of erosion rate estimates. For example, temporally increasing erosion rates produce poor agreement between erosion rate estimates made with a single method, while decreasing erosion rate produces good agreement between estimates made with a single method. Nonuniform erosion that decreases relief can produce erroneously high erosion rate estimates based on the elevation dependence method but also accurate and relatively precise mineral pair-based estimates. A history of decreasing relief can be distinguished from one of steady, uniform erosion by (1) close agreement between mineral pair estimates and poor agreement between elevation dependence estimates and (2) the magnitude of the difference between FT and He ages. The insights gleaned from these analyses can guide forward modeling of topographic and thermal evolution.