Abstract We present 17 new 10Be erosion rates from southern Peru sampled across an extreme orographic rainfall gradient. Using a rainfall-weighted variant of the normalized channel steepness index, ksnQ, we show that channel steepness values, and thus topography, are adjusted to spatially varying rainfall. Rocks with similar physical characteristics define distinct relationships between ksnQ and erosion rate (E), suggesting ksnQ is also resolving lithologic variations in erodibility. However, substantial uncertainty exists in parameters describing these relationships. By combining our new data with 38 published erosion rates from Peru and Bolivia, we collapse the range of compatible parameter values and resolve robust, nonlinear ksnQ–E relationships suggestive of important influences of erosional thresholds, rock properties, sediment characteristics, and temporal runoff variability. In contrast, neither climatic nor lithologic effects are clear using the traditional channel steepness metric, ksn. Our results highlight that accounting for spatial rainfall variations is essential for disentangling the multiple influences of climate, lithology, and tectonics common in mountain landscapes, which is a necessary first step toward greater understanding of how these landscapes evolve.