This paper describes a cost-effective technique that optimally utilizes all available diagnostic studies for three-dimensional treatment planning. A simulator unit modified to produce cross-sectional images (simulator-CT unit) is used to create a reference data set with the patient in the treatment position. Registration software (qsh) brings other diagnostic studies into agreement with this reference data set. Two cases are presented as examples of the use of this technique. Registration of abdominal scans from the same patient demonstrates the warping of a nontreatment position study to the treatment position. The second case is based on paired data sets through the head, in which the diagnostic study was obtained by using a gantry tilt to follow the base of the skull and to avoid sections passing through the teeth. The registration software provides a method for combining diagnostic studies into a single "master" data set. The success of the transformation depends on the operator's ability to identify corresponding anatomic landmarks for different data sets and on the magnitude of the variation in the patient's position from one procedure to the next. Limitations in image quality and the number of cross-sections obtainable from a simulator-CT unit can be partially overcome by using the described technique. Thus, the information contained in nontreatment position diagnostic tests can be used accurately for treatment planning at limited cost.