A "low-cost" personal computer (PC) system used to digitize dental radiographs was tested by assessing the accuracy of its subtraction images versus those of "high-cost" industrial equipment and conventional radiography. Subtraction images were made of artificial lesions in human femur bone and subsequently evaluated by students and teachers. The observations were analyzed in terms of true positive and false positive reports. "Low-cost" and "high-cost" subtraction images revealed only small differences in diagnostic accuracy. Compared to conventional radiography, the diagnostic accuracy of the subtraction images with the "low-cost" PC system was significantly higher for all observers. The interexaminer variance was similar for the subtraction and the conventional images for both students and teachers, except for a significantly reduced interexaminer variance for the teachers concerning the true positive reports with the "low-cost" PC subtraction technique.