We have tested a network of magnetic direction-finders (DFs) that locate ground strikes in Oklahoma and surrounding states in order to determine detection efficiency for the network and systematic errors in azimuth (i.e., site errors) for each of four DF sites. Independent data on lightning strike locations were obtained with a television (TV) camera on a mobile laboratory and an all-azimuth TV system at the National Severe Storms Laboratory (NSSL). In two tests using these data, we found a location detection efficiency of about 70% for storms at about 70 and 300 km from the center of the network. Systematic errors in azimuth were determined by comparing locations from the lightning strike locating system with strikes located from the mobile laboratory system; also, for a single DF at NSSL, strike azimuths from the DF were compared with azimuths from the all-azimuth TV system for storms near NSSL. Furthermore, we developed a technique for using redundant DF data to determine systematic errors in azimuth measurements for each DF site. Azimuthal errors found by this analytic technique were consistent with errors found by using the two sets of direct measurements. The azimuthal errors are themselves a function of azimuth, with peak amplitudes ranging from less than 5° for DFs located at favorable sites to about 11° for one DF located at an unfavorable site.