Constraining Earth’s sediment mass balance over geologic time requires a quantitative understanding of how landscapes respond to transient tectonic perturbations. However, the mechanisms by which bedrock lithology governs landscape response remain poorly understood. Rock type influences the size of sediment delivered to river channels, which controls how efficiently rivers respond to tectonic forcing. The Mendocino triple junction region of northern California, USA, is one landscape in which large boulders, delivered by hillslope failures to channels, may alter the pace of landscape response to a pulse of rock uplift. Boulders frequently delivered by earthflows in one lithology, the Franciscan mélange, have been hypothesized to steepen channels and slow river response to rock uplift, helping to preserve high-elevation, low-relief topography. Channels in other units (the Coastal Belt and the Franciscan schist) may experience little or no erosion inhibition due to boulder delivery. Here we investigate spatial patterns in channel steepness, an indicator of erosion resistance, and how it varies between mélange and non-mélange channels. We then ask whether lithologically controlled boulder delivery to rivers is a possible cause of steepness variations. We find that mélange channels are steeper than Coastal Belt channels but not steeper than schist channels. Though channels in all units steepen with increasing proximity to mapped hillslope failures, absolute steepness values near failures are much higher (∼2×) in the mélange and schist than in Coastal Belt units. This could reflect reduced rock erodibility or increased erosion rates in the mélange and schist, or disproportionate steepening due to enhanced boulder delivery by hillslope failures in those units. To investigate the possible influence of lithology-dependent boulder delivery, we map boulders at failure toes in the three units. We find that boulder size, frequency, and concentration are greatest in mélange channels and that Coastal Belt channels have the lowest concentrations. Using our field data to parameterize a mathematical model for channel slope response to boulder delivery, we find that the modeled influence of boulders in the mélange could be strong enough to account for some observed differences in channel steepness between lithologies. At the landscape scale, we lack the data to fully disentangle boulder-induced steepening from that due to spatially varying erosion rates and in situ rock erodibility. However, our boulder mapping and modeling results suggest that lithology-dependent boulder delivery to channels could retard landscape adjustment to tectonic forcing in the mélange and potentially also in the schist. Boulder delivery may modulate landscape response to tectonics and help preserve high-elevation, low-relief topography at the Mendocino triple junction and elsewhere.