The importance of circulating blood (BV) and plasma volume (PV) in critically ill patients and physiological research is unchallenged. Recently, Evans blue (EB) [8, 25] and radioactively labelled serum albumin (RIHSA) [20] have mostly been used as tracers for PV determination. However, the disadvantages of radioactive contamination (RIHSA) and dye accumulation (EB), especially in repeated measurements, are obvious. In addition, recent reports show a possible carcinogenic potential for EB [15, 21]. This has prompted us to examine the feasibility of indocyanine green (ICG), a tricarbocyanine dye currently used for cardiac output and liver blood flow measurements, for the determination of PV. The volume of distribution of ICG has been reported to represent PV [5, 26]. METHODS. In 23 healthy volunteers (19 men and 4 women), PV was determined in duplicate (PV1, PV2) with an interval of 30 min. Before injection a tourniquet was put around the arm and a pressure above the systolic arterial pressure was applied for 2 min. During recirculation, ICG (2.5 mg/ml) was administered in a dose of 0.25 mg/kg as a bolus injection over 5 s via an antecubital vein. Blood was drawn from an antecubital vein of the contralateral arm at 1 min intervals. After centrifugation, the optical density (corrected for blank) was read in a densitometer. Third- to ninth-minute plasma samples were used to calculate monoexponential plasma decay curves. The ICG concentration at injection time was achieved by extrapolation. A calibration curve was generated using 5 different known ICG concentrations. PV was calculated from injected ICG dose divided by ICG concentration at injection time. BV and red cell volumes (EV) were derived from measured PV and hematocrit (hct). RESULTS. Between minutes 3 and 9, tracer decay was monoexponential in all but 1 subject. From minute 10 on the plasma decay of ICG represented another, slower compartment (Fig. 1). The plasma half-life of ICG was 3.2 +/- 0.6 min (mean +/- SD). Mean PVs per body weight and body surface area (BSA) were 44 +/- 5 ml/kg and 1662 +/- 176 ml/m2, respectively. Linear regression revealed PV2 = 0.92.PV1 + 226 (r = 0.92) (Fig. 2). The mean percentage of difference (D) was -0.6%, the methodologic error (SD) +/- 5.7% [27]. Linear regression of PV and BSA revealed PV = 1885.BSA -416 (r = 0.71, P less than 0.0001) (Fig. 3). BV and EV estimates (Table 2) obtained from PV and hct showed reproducibility in the range of the PV determination because of excellent reproducibility of hct measurements. DISCUSSION. ICG plasma half-life times in our experiments were comparable to those reported by other authors [18, 19, 24]. Reproducibility of PV determination was good and was well within the limits of other tracer methods (EB, RIHSA) [17, 27]. Using exclusively peripheral veins for ICG injection and blood withdrawal did not seem to affect the accuracy of PV determination. PV estimates obtained by the ICG method showed good agreement with those known from the literature [7, 10, 25]. Our results correspond especially well with the data reported by Hurley [14] obtained from 481 healthy men using different methods (Evans blue, RIHSA, or labelled red cells).