Canids can be difficult to detect and their populations difficult to monitor. We tested whether hair samples could be collected from coyotes (Canis latrans) in Texas, USA and gray wolves (C. lupus) in Montana, USA using lure to elicit rubbing behavior at both man-made and natural collection devices. We used mitochondrial and nuclear DNA to determine whether collected hair samples were from coyote, wolf, or nontarget species. Both coyotes and wolves rubbed on man-made barbed surfaces but coyotes in Texas seldom rubbed on hanging barbed surfaces. Wolves in Montana showed a tendency to rub at stations where natural-material collection devices (sticks and debris) were present. Time to detection was relatively short (5 nights and 4 nights for coyotes and wolves, respectively) with nontarget and unknown species comprising approximately 26% of the detections in both locations. Eliciting rubbing behavior from coyotes and wolves using lures has advantages over opportunistic genetic sampling methods (e.g., scat transects) because it elicits a behavior that deposits a hair sample at a fixed sampling location, thereby increasing the efficiency of sampling for these canids. Hair samples from rub stations could be used to provide estimates of abundance, measures of genetic diversity and health, and detection-nondetection data useful for cost-effective population monitoring.