Deadtime performance of scintillation cameras is sensitive to such factors as scatter fraction and analyzer window width. Data from manufacturers and previous investigators do not predict counting-rate losses under clinical conditions. Scintillation cameras used with Tc-99m for quantitative nuclear cardiology should be evaluated for deadtime performance by the two-source method using a scatter phantom designed to simulate the spectrum from Tc-99m in the heart. Under these conditions, scintillation cameras were found to follow the paralyzable model; accurate estimates could be obtained for data losses and maximum useful counting rates in a clinical setting. A survey of 39 contemporary scintillation cameras yielded a range of paralyzing deadtime values of 4.3 to 10 musec, with a 20% window centered on the Tc-99m photopeak. For an average deadtime of 6 musec, counting rates should be maintained below 36,000 cps to avoid undue data losses in excess of 25%.