Documented evidence indicates that military training lands support greater diversity of threatened and endangered species than lands managed by other government agencies. The intermediate disturbance hypothesis suggests that biodiversity is maximized where the frequency and/or intensity of disturbance occurs at an intermediate level.  While the logic of the hypothesis is sound, and has been demonstrated to have validity in many cases, a missing component of the hypothesis is the diversity of the disturbance itself.  Numerous ecologists have suggested that disturbance may create patchiness of habitats that, in turn, contributes to increased biodiversity at the landscape scale. Based on quantitative evidence and personal observations from military training lands, I have proposed a heterogeneous disturbance hypothesis suggesting that where multiple kinds, frequencies, severities, and durations of disturbance occur concomitantly in a spatially and temporally distributed fashion, biodiversity will be greater than where any single level of disturbance is applied uniformly across the landscape.  To test the heterogeneous disturbance hypothesis, I collected data from Hohenfels Training Area, Germany. Within a hillslope with relatively uniform soil and aspect, I recorded plant species richness in twenty 1m² plots in each of several conditions along a disturbance continuum ranging from heavily disturbed to nearly pristine. The data were evaluated using a statistical technique called ‘bootstrapping’.  Repeated sub-samples were randomly taken from plots within each disturbance condition and compared with an equivalent number of sub-samples randomly taken from across all disturbance conditions. Species richness was significantly greater when samples were taken across disturbance conditions than when taken within any single disturbance condition.  The data support the heterogeneous disturbance hypothesis and help explain the enhanced biodiversity that characterizes military training areas.