The diagnostic accuracy of realtime teledermatology was measured using two different video cameras. One camera was a relatively low-cost, single-chip device (camera 1), while the other was a more expensive, three-chip camera (camera 2). The diagnosis obtained via the videolink was compared with the diagnosis made in person. Sixty-five new patients referred to a dermatology clinic were examined using camera 1 followed by a standard face-to-face consultation on the same day. A further 65 patients were examined using camera 2 and the same procedure implemented. Seventy-six per cent of conditions were correctly diagnosed by telemedicine using camera 2 compared with 62% using camera 1. A working differential diagnosis was obtained in 12% of cases using camera 2 compared with 14% using camera 1. The percentage of 'no diagnosis', wrong and missed diagnoses was halved using camera 2 compared with camera 1. These results suggest that the performance of the more expensive camera was superior for realtime teledermatology.