The inability to measure borders and variation in the number of 1-mm cross sections acquired from an identical length of vessel in serial intravascular ultrasound (IVUS) pullbacks represents potential errors in calculating volumes by IVUS. In a clinical IVUS trial, the percentage of nonmeasurable lumen and external elastic membrane borders, and the percent variation in the number of 1-mm cross sections acquired from an identical vessel length at 2 separate time points, were determined. A statistical model that simulated the effect of varying the percentage of the total number of cross sections in a pullback (i.e., sample fraction) was developed using SAS software. Mean and maximum errors for calculation of atheroma volume for each sample fraction were determined. The mean percentage of nonmeasurable lumen and external elastic membrane borders in an individual patient was 8.4 +/- 8.4% and 17.4 +/- 18.4%, respectively. Mean variation in the number of 1-mm cross sections acquired in serial studies was 5.6 +/- 6.2%. A decrease in sample fraction from 95% to 50% was associated with a linear increase in the mean and maximum errors in atheroma volume, from 2.0 +/- 0.9% and 5.9 +/- 3.0%, to 7.1 +/- 2.8% and 23.4 +/- 10.3%, respectively. Thus, nonmeasurable borders and variation in the number of 1-mm cross sections acquired from an identical length of vessel in serial studies are real considerations in clinical IVUS trials. However, given the reported incidence of these considerations in this clinical trial, our statistical model suggests that the impact of each of these considerations on atheroma volume calculation is small.