Serial determination of radionuclide left ventricular ejection fraction and volumes require a label that remains highly stable after introduction into the human circulation. We have evaluated the in vivo stability of 99Tcm-red blood cells (99Tcm-RBC) labelled in vitro by preparation with small amounts of a stannous agent in 19 patients with coronary artery disease. The distribution volume of 99Tcm-RBC was virtually identical to that of RBC labelled with 51Cr. Labelling efficiency expressed as the cell bound/total activity was more than 95% and remained high throughout 1 h after injection of the label. The effective in vivo half-times (T1/2) calculated from venous blood activity and externally recorded left ventricular end-diastolic frame activity were 342 +/- 103 min (mean +/- S.D.) and 306 +/- 92 min, respectively. A significant correlation, r = 0.86 (p less than 0.01) was found between the T1/2 values calculated from the two methods. Thus, a high in vivo stability of the label was demonstrated with considerable inter-patient variation. The labelling procedure seems suitable for serial performance of radionuclide cardiography within an hour after injection of 99Tcm-RBC. However, serial volume determination necessitates individual calculation of in vivo tracer decay from either venous blood or externally recorded activity in the left ventricular area during steady conditions.