Modern computed tomographic (CT) scanners allow reasonably high-resolution cross-sectional visualization of most viscera and masses. To determine the accuracy of CT volume estimation, CT volumes of inanimate objects, cadaver kidneys and spleens, and in vivo balloons were performed. Regions of interest were outlined by a hand-operated cursor, and the computer program calculated the cross-sectional area in square millimeters and in pixels. A second computer program used Simpson's rule to calculate the volume from these multiple cross-sectional areas. Calculated CT volumes were within ±10% of directly measured volumes. Tumor masses being treated by chemotherapy were followed up clinically and by CT. Volume changes determined by CT are believed to be equal to and frequently more sensitive than clinical examination. From our experimental CT and clinical experiences, accuracy can be affected by (1) respiratory movement; (2) rapid changes in in vivo blood volume; (3) low CT number-gradient at the objects' periphery; (4) observer error in cursor tracing of the desired structure; and (5) mathematical errors inherent in Simpson's rule. We conclude that CT can estimate volumes of CT definable masses and can be useful in following tumor response to therapy.