Disturbance frequency, intensity, and areal extent may influence the effects of disturbance on biological communities. Furthermore, these three factors may have interacting effects on biological diversity. We manipulated the frequency, intensity, and area of disturbance in a full-factorial design on artificial substrates and measured responses of benthic macroinvertebrates in a northern Vermont stream. Macroinvertebrate abundance was lower in all disturbance treatments than in the undisturbed control. As in most other studies in streams, species density (number of species/sample) was lower in disturbed treatments than in undisturbed controls. However, species density is very sensitive to total abundance of a sample, which is usually reduced by disturbance. We used a rarefaction method to compare species richness based on an equivalent number of individuals. In rarefied samples, species richness was higher in all eight disturbed treatments than in the undisturbed control, with significant increases in species richness for larger areas and greater intensities of disturbance. Increases in species richness in response to disturbance were consistent within patches, among patches with similar disturbance histories, and among patches with differing disturbance histories. These results provide some support for Huston's dynamic-equilibrium model but do not support the intermediate-disturbance hypothesis. Our analyses demonstrate that species richness and species density can generate opposite patterns of community response to disturbance. The interplay of abundance, species richness, and species density has been neglected in previous tests of disturbance models.