I studied the characteristics of fence mortality in pronghorn (Antilocapra americana), mule deer (Odocoileus hemionus), and elk (Cervus elaphus) along roads in Colorado and Utah from June 2004 to June 2005. I defined a direct-fence mortality as a carcass caught directly in a fence and an indirect-fence mortality as a carcass on the ground within 10 m of a fence. I estimated an average annual direct mortality occurrence of 0.25 mortalities/km (0.078 mule deer mortalities/km, 0.113 pronghorn mortalities/km, and 0.061 elk mortalities/km). The highest fence-mortality rates for ungulates occurred during August, which coincided with weaning of fawns on my study area. Mule deer and pronghorn both jumped fences in >81% of observed crossings and did not differ in their crossing methods (P = 0.37). Getting caught between the top 2 wires was the leading cause of death for fence mortalities. Mule deer suffered higher fence-mortality rates than elk or pronghorn because they crossed fences more frequently and fed in the right-of-way of the road more often (P < 0.001). Juveniles were 8 times more likely to die in fences than adults. Woven-wire fence types were more lethal to ungulates (especially juveniles) than other fence types (P < 0.001). Woven wire with a single strand of barbed wire above it was significantly more lethal to ungulates than woven wire with 2 strands of barbed wire above it, or 4-strand barbed-wire fence (P < 0.001). There was a direct relationship between the frequency of fence-mortalities and ungulate abundance (P < 0.001 ). Traffic volumes had an inverse relationship with fence mortality frequencies (P < 0.001) and ungulate densities along the right-of-way (P < 0.001). Indirect mortality (i.e., carcasses within 10 m of fences) composed 66% of fence-related mortality, whereas direct-fence mortality (i.e., carcasses in fences) composed a mere 33%. Additionally, indirect-fence mortality was found to be greater along woven-wire fences, when compared to barbed-wire fence types (P = 0.003).