Purpose: To investigate the dosimetric differences between colpostats and cylinder applicators for intravaginal brachytherapy. Methods and Materials: Dose distributions near vaginal colpostats and dome cylinders were computed with a commercial high-dose-rate treatment-planning system and verified by spot measurements by using LiF thermoluminescent dosimeters. Taking source anisotropy into account, dwell times were optimized by the computer by using the polynomial optimization on dose points method to give uniform doses along the lateral surfaces of the applicators. In addition, the effects of vaginal packing and the separation distance between colpostats were studied by computing the dose to the vaginal mucosa, assuming 0.5 and 1.0 cm of vaginal packing and colpostat separation, respectively. Results: The computed and measured doses agreed within ±7%. Surface doses were similar for both types of applicators when the effect of shielding in the colpostats was neglected. However, percent depth doses in the anterior/posterior and lateral directions were higher for the cylinder, whereas the dose fall-off along the longitudinal patient axis was less pronounced for the colpostats. When vaginal packing at the anterior and posterior surface of the colpostats was increased from 0 to 5, 10, and 15 mm, the corresponding vaginal dose decreased from 97% of the prescription dose to 60%, 39%, and 26%, respectively. Separating the colpostats from 0 to 5 and 10 mm reduced the surface dose near the bladder/rectum to 80% and 67%, respectively, whereas the respective apex dose decreased from 105% of prescription to 91% and 77%. Conclusions: Colpostats and cylinder applicators for intracavitary brachytherapy have their advantages and disadvantages in depth dose distribution and clinical use. If treatment is confined to the vaginal apex, either applicator can be used. However, the colpostat separation should be kept to a minimum, and vaginal packing should be applied with great care to avoid generating cold spots along the upper vaginal surface and vaginal cuff. © 2002 Elsevier Science Inc. All rights reserved.