Recent reports have questioned the accuracy of the indocyanine green dilution technique for measuring plasma volume. Our objective was to evaluate the impact of different time windows for monoexponential extrapolation. We retrospectively analysed 31 indocyanine green decay curves to investigate the problem in principle (group 1) and prospectively performed another 21 plasma volume measurements to estimate its practical impact (group 2). To monoexponentially extrapolate back to the specific extinction at the time of dye injection, two different time windows were applied to each decay curve, comparing the plasma volumes resulting from sampling within a short (<or=5 min) versus a longer (>5 min) period of time. Extrapolating back from the longer period led to a higher apparent plasma volume relative to the shorter period in both groups, the difference being 348 +/- 171 ml (group 1) and 384 +/- 131 ml (group 2; mean +/- SD; p < 0.05 each). This result was due to a reliable monoexponentiality of decay only up to the 5th min after dye injection. Thus, to estimate the initial distribution space of indocyanine green via monoexponential extrapolation, the first linear kinetic of indocyanine green decay should be taken.