Direct and indirect contact between wild and farmed cervids along perimeter fences may play a role in transmission of diseases like chronic wasting disease (CWD), but no studies have quantified such interactions. At 9 high-fenced commercial elk (Cervus elaphus) farms in Colorado, USA, during October 2003 to January 2005, we used animal-activated video to estimate rates of fence-line use by wild cervids, rates of direct contact between farmed and wild cervids, and probability of direct contact when wild cervids were present. We recorded 8-fold-more wild elk per unit time than mule deer (Odocoileus hemionus) at fence lines. Depending on site, we recorded 0.66 to 46.90 wild elk per 1,000 hours of camera monitoring. We documented 77 interactions between wild and farmed elk involving naso-oral contact and no contact between wild mule deer and farmed elk. Rate of direct contact ranged from 0.00 to 1.92 direct contacts per 1,000 hours of camera monitoring among sites. Given recorded presence of wild elk, estimated probability of observing direct contact during a 2-minute video recording ranged from 0.00 to 0.11 among sites. Risk of direct contact was about 3.5 times greater for single woven-wire fence compared with offset electric fence attached to a single woven-wire fence. We observed no direct contact through double woven-wire fences. Because little is currently known about infection rates associated with infection mechanisms, we cannot infer a level of CWD infection risk from our results, but some form of double fencing should reduce potential for direct and indirect transmission of disease into or out of elk farms.