Three radiologists examined 34 venograms (68 limbs) performed supine without fluoroscopy for evidence of thrombosis. At 11 major veins in each limb they stated whether thrombus was "absent," "doubtful," "presumed" or "definite" or expressed " "no opinion." Data were analyzed by using the kappa statistic, which corrects for chance agreement. Results were compared with those from a previous study in which two of the observers assessed venograms performed under fluoroscopic control with the patient semi-erect. In the present study observers expressed "no opinion" much more often, particularly for anterior tibial and iliac veins. With these examinations excluded, the extent of observer variation was similar in the two studies, disagreement about the presence of thrombus occurring in about 10% of examinations. The frequency with which the diagnosis of thrombus remains in doubt severely limits the value of venography performed supine without fluoroscopy. Fluoroscopically-controlled examinations in the semi-erect position are preferable.