Simple measures of habitat proximity made primarily on the basis of land cover are widely used in the ecological literature to infer habitat connectivity, or the potential for animal movement among resource patches. However, such indices rarely have been tested against observations of animal movement or against more detailed biological models. We developed a priori expectations as to the types of study systems and organisms for which various habitat proximity indices would be best suited. We then used data from three study systems and four species to test which, if any, of the indices were good predictors of population-level responses. Our a priori expectations about index performance were not upheld. The indices that consider both habitat area and distance from the focal patch were highly correlated with each other, suggesting that they do index similar quantities. However, none of the indices performed well in predicting population response variables. The results suggest that the pattern of habitat cover alone may be insufficient to predict the process of animal movement.