The findings of this study provide preliminary evidence to suggest that the first in a series of three thermodilution cardiac output measurements is statistically significantly higher than the second and third measurements when 10 ml of iced 5% dextrose is used as the injectate. The repeated measures analysis of variance was highly significant (p = 0.00001) for the sample (N = 61) and for a subset of patients (N = 34) with concurrent central infusions of 1 to 25 ml/hr (p = 0.004). The first measurement was the highest reading in the series more frequently than expected (54% of the sample). However 23% of the subjects had the second measurement as the highest reading and 22% had the third reading as the highest measurement in the series. The effect of the concurrent central infusions on this set of data is not known. The mean difference between sequential measurements was smallest between the second and third measurements (0.33 L/min) but the mean difference between the first and second measurement (0.38 L/min) and the first and third measurement (0.44 L/min) was of similar magnitude. The first measurement differed from the average of the second and third measurements by more than 10% or by greater than 0.5 L/min in 26% (N = 16) of the readings for the entire sample; this figure was 38% (N = 13) in the subset of patients with concurrent central infusions of 1 to 25 ml/hr. It is important that the measurement protocol be consistent and that triplicate measurements be done so that trends in a patient's hemodynamics can be accurately assessed and the effects of therapeutic interventions evaluated. Data should be cautiously interpreted if the first measurement is significantly higher than the second and third measurements because hemodynamically compromised patients could be adversely affected by this type of technical error.