The technical basis for the dose estimates for several radiopharmaceuticals used in nuclear cardiology is reviewed, and cases in which uncertainty has been encountered in the dosimetry of an agent are discussed. Also discussed is the issue of uncertainties in radiation dose estimates and how to compare the relative risks of studies. Methods: Radiation dose estimates (organ absorbed doses and effective doses) from different literature sources were directly compared. Typical values for administered activity per study were used to compare doses that are to be expected in clinical applications. Results: The effective doses for all agents varied from 2 to 15 mSv per study, with the lowest values being seen for 13 N-NH 3 and 15 O-H 2 O studies and the highest values being seen for 201 Tl-chloride studies. The effective doses for 99m Tl-labeled agents differed by about a factor of 2, a factor that is comparable to the uncertainty in individual values. This uncertainty results from the application of standard anthropomorphic and biokinetic models, presumably representative of the exposed population, to individual patients. Conclusion: Considerations such as diagnostic accuracy, ease of use, image quality, and patient comfort and convenience should generally dictate the choice of a radiopharmaceutical, with radiation dose being only a secondary or even tertiary consideration. Counseling of nuclear medicine patients who may be concerned about exposure should include a reasonable estimate of the median dose for the type of examination and administered activity of the radiopharmaceutical; in addition, it should be explained that the theoretic risks of the procedure are orders of magnitude lower than the actual benefits of the examination. Providing numeric estimates of risks from studies to individual patients is inappropriate, given the uncertainties in the dose estimates and the limited predictive power of current dose-risk models in the low-dose (i.e., diagnostic) range.