We used a suite of topographic metrics and cosmogenic 10Be-derived catchment-averaged denudation rates from 18 watersheds to evaluate patterns of millennial-scale erosion of the Santa Lucia Mountains and test the explanatory power of the power-law incision rule for this landscape. Catchment-averaged denudation rates in the study area vary between ∼0.07 and 0.4 mm/yr, with a single drainage yielding a rate >0.45 mm/yr. Channel steepness ranges from ∼90 to 390 m for these catchments. We used these observations to test two forms of the power-law incision rule, one incorporating multiple, lithologically dependent erodibilities, and one containing a single erodibility term. Statistical analyses indicated the power-law incision rule provides an improved fit relative to a model that does not relate erosion rate to channel steepness. However, tests comparing both model forms indicate no significant improvement in model fit when using unique erodibility and channel steepness terms for each lithology. Estimates of erodibility coefficients were ∼10–6–10–5 m0.2/yr for all rock types evaluated. Where range asymmetry was most pronounced, watershed divides were identified as mobile by the χ landscape metric, yet we resolved no cross-divide change in erosion rate. Conversely, divides predicted by χ to be close to an equilibrium configuration showed markedly different erosion rates in adjoining basins. These conflicting data sets, considered in conjunction with geomorphic indicators of river capture, may suggest that divide migration occurs in abrupt, punctuated episodes, rather than by gradual migration of divides. Finally, erosion rates measured here are broadly consistent with late Pleistocene uplift rates recorded by geomorphic markers and are a factor of 2–6 lower than previously reported rates of exhumation gauged by apatite helium thermochronology. This suggests that there has been a secular slowing of erosion, and perhaps rock uplift rates, between the Pliocene and today.