The accuracy of microhiitological techniques for analysis of herbivore diets was evaluated with cattle, sheep, and Angora goats fed grass, forb, and shrub mixtures of known botanical compositions. Two observers performed microhistological analyses on undigested diets as offered and on feces collected. Similarity indices and chi-square tests were used to determine if differences existed among actual diets, estimated diets, and fecal samples. Botanical compositions of diets fed to all 3 anhnal species generally were accurately estimated by fecal analyses. In some other studies, shrubs in ruminant diets have been inaccurately estimated by the microhistological technique. However, in our study, shrubs were accurately estimated with no differences between actual and observed compositions. We attribute this to the fact that shrub materials used in our study had a high proportion of current growth relative to woody materials. Woody plant parts had lower proportions of identifiable epldermal material than leaves and young stems. In grass-forb diets, forbs sometimes were overestimated and differentiation among grasses was difficult. However, In most cases, observers could precisely estimate diets of the 3 herbivore species.