Characterizing habitat features that influence beaver (Castor canadensis) occupancy along roadsides may have important implications for managing damage to roads caused by beaver activity. We initiated this study to develop proactive and long-term approaches to deal with nuisance beaver along roadsides. From June to October 1997 and 1998, we sampled 316 roadside sites in New York state, USA—216 sites where beaver occupied the roadside area and 100 unoccupied sites. We used stepwise logistic regression to identify habitat variables associated with beaver occupancy along roadsides. We evaluated regression models through measures of sensitivity and specificity. The logistic function retained the percentage of roadside area devoid of woody vegetation, stream gradient, the interaction between these 2 variables, and stream width in the final model. Precluding beaver occupancy along highways would necessarily involve large-scale removal of woody vegetation that would be impractical in all but the most intensive management scenarios. However, beaver habitat assessment adjacent to roads may be a useful tool for designing new highways, prioritizing culvert replacements, and developing proactive plans for beaver damage management.