The use of an indicator dye for serial determinations of cardiac function by the dye dilution technic requires in vitro stability of the dye for the duration of the study period, since repeated calibration curves are not routinely performed. The light absorption of a commonly used tricarbocyanine dye (Cardio Green) in the manufacturer's solvent and in whole blood was studied continuously (Beckman DK-2 spectrophotometer) for 8 hours following preparation of the dye solution. Although the wave length of peak absorbance in aqueous solution did not deviate from 775 millimicrons, a progressive increase in absorbance (as much as 10% in 8 hours) was consistently observed, in direct contrast to previous reports. However, throughout the period of study, the dye continued to conform to the Beer-Lambert Law, thus permitting correction of the changing absorbance by construction of a calibration curve each time the dye is used. Unless such a correction is made, a significant error can be introduced into serial estimations of cardiac output.