Biological invasion is a widespread, but poorly understood phenomenon. Elton's hypothesis, supported by theory, experiment, and anecdotal evidence, suggests that an important determinant of invasion success is resident biodiversity, arguing that high diversity increases the competitive environment of communities and makes them more difficult to invade. Observational studies of plant invasions, however, find little support for this hypothesis and argue strongly against it. Lack of control of extrinsic factors (e.g., disturbance, climate, or soil fertility) that covary with biodiversity and invasion in observational studies makes it difficult to determine if their findings truly refute Elton's hypothesis. We examined performance of Crepis tectorum (an invasive, annual composite weed) in experimental prairie grassland plots and greenhouse plant assemblages in which resident species richness was directly manipulated. Under these conditions, unlike observational studies, no covarying extrinsic factors could interfere with interpreting results. We found a strong inverse association between resident diversity and invader performance as predicted by Elton's hypothesis. Higher resident diversity increased crowding, decreased available light, and decreased available nutrients all of which increased the competitive environment of diverse plant assemblages and reduced C. tectorum success. Examination of individual resident species impacts on C. tectorum performance demonstrated that this diversity effect was not due to the sampling effect. These results suggest that both Elton's hypothesis and its competitive mechanism may operate in nature, but covarying extrinsic factors may obscure the negative impact of diversity on invader success.