Woody plant encroachment is known to have substantial effects on a range of ecosystem processes. Research worldwide indicates that the area around shrubs and trees has higher levels of infiltration than the interspaces. Little is known, however, about the hydrological consequences of shrub removal on infiltration, and how this might be influenced by grazing. We examined the spatial patterns of infiltration across three treatments relating to shrub removal and grazing: (i) undisturbed (ungrazed, unploughed), (ii) grazed but unploughed, and (iii) grazed and ploughed. In general we found that disturbance was associated with a greater cover of bare soil but lower infiltrability, our laboratory-based measure of infiltration. At the undisturbed site, bare soil was patchy and localized, with an autocorrelation range or connectivity of 1.4 m. The autocorrelation range of infiltrability at this site (A0 = 3 m) was larger than would be predicted from the size of the shrub canopy, and this was attributed to the presence of a well-developed understorey layer and biological soil crust community. At both grazed sites, infiltration was confined to the immediate canopy area of the remaining shrubs (A0 = 1.2 m in the unploughed-grazed site). Additionally, there was increasing connectivity of bare soil with disturbance, up to 6.8 m at the ploughed-grazed site. With increasing disturbance, resource-rich shrub patches are likely to become more developed, further reinforcing their growth and persistence at the expense of the bare interspaces. Our results indicate the importance of shrubs for maintaining landscape connectivity, and the long-term unsustainable practice of removal by ploughing, which is likely to promote shrub dominance rather than suppression.