This study prospectively examined x-rays obtained in the offices of 14 family physicians who read all their own office x-rays and then select those they want a radiologist to read. Each physician completed a standardized report form for 100 consecutive x-rays or for all x-rays obtained for three months, whichever came first. A total of 1,127 x-rays were evaluated. The results indicate that: 1) 86.5% of all x-rays were of three types (chest x-ray, 50.8%; upper extremity, 18.9%; lower extremity, 16.8%); 2) 85.3% of all x-rays were obtained to: a) confirm a clinical diagnosis, b) establish a diagnosis, c) screen for potential problems; 3) 84.8% of the referred x-rays were referred to a) confirm the family physician's interpretation, b) interpret equivocal findings, c) provide medicolegal support, d) improve interpretation skills; 4) the referral rate to a radiologist was 11.6%, with chest x-rays referred most frequently; 5) the same five major ICD-9-CM diagnostic categories were associated with 90.1% of all x-rays obtained and 86.5% of all x-rays referred; 6) the types of x-rays obtained in community practice and in a previously studied residency training practice were similar; 7) 55.6% of office x-rays were read as "normal;" and 8) the family physicians' discordancy rate with the radiologists on referred films was 12.5%.