A systematic investigation was undertaken to determine which factors affect the calibration of a Gilford monochromatic densitometer used for recording indicator-dilution curves with indocyanine green. Calibration was not influenced by variation of withdrawal speeds from 6.4 to 20.6 cm3/min but varied with differing sampling systems. Variation in haemoglobin from 10 to 17.5 g-dl-1 or in haematocrit from 30 to 60% did not affect calibration. Calibration lines showed both curvature and hysteresis. The opitcal properties of indocyanine green were not affected by storage of diluted dye in the dark for 24 hours but were altered by the presence of contrast media used in angiography. No systematic error resulted from the substitution of stored ACD blood for patient's blood in calibration although random error was increased. Recommendations based on these findings are made for a simple calibration technique which avoids systematic error.