In semiarid landscapes, the linkage between runoff and vegetation is a particularly close one. In this paper we report on the results of a long-term and multiple-scale study of interactions between runoff, erosion, and vegetation in a piñon–juniper woodland in New Mexico. We use our results to address three knowledge gaps: (1) the temporal scaling relationships between precipitation and runoff; (2) the effects of spatial scale on runoff and erosion, as influenced by vegetation; and (3) the influence of disturbance on these relationships. On the basis of our results, we tested three assumptions that represent current thinking in these areas (as evidenced, for example, by explicit or implicit assumptions embedded in commonly used models). The first assumption, that aggregated precipitation can be used as a surrogate for total runoff in semiarid environments, was not verified by our findings. We found that when runoff is generated mainly by overland flow in these systems, aggregated precipitation amounts alone (by year, season, or individual event) are a poor predictor of runoff amounts. The second assumption, that at the hillslope and smaller scales runoff and erosion are independent of spatial scale, was likewise not verified. We found that the redistribution of water and sediment within the hillslope was substantial and that there was a strong and nonlinear reduction in unit-area runoff and erosion with increasing scale (our scales were slope lengths ranging from 1 m to 105 m). The third assumption, that disturbance-related increases in runoff and erosion remain constant with time, was partially verified. We found that for low-slope-gradient sites, disturbance led to accelerated runoff and erosion, and these conditions may persist for a decade or longer. On the basis of our findings, we further suggest that (a) disturbance alters the effects of scale on runoff and erosion in a predictable way—scale relationships in degraded areas will be fundamentally different from those in nondegraded areas because more runoff will escape off site and erosion rates will be much higher; and (b) there exists a slope threshold, below which semiarid landscapes will eventually recover following disturbance and above which there will be no recovery without mitigation or remediation.