Accurate and reproducible measures are required to study arterial stiffness in human populations. The reproducibility of aortic pulse wave velocity was evaluated in 14 participants from a population-based study of cardiovascular disease in the elderly. Three data files were collected per participant by each of two sonographers and files were read by two readers. Seven of the 14 participants returned for a second visit 1 week later to assess between-visit variability. Reproducibility was evaluated with Pearson and intraclass correlations and by the absolute value of the difference between replicate values. The overall reliability coefficient was rI = 0.77. Between-sonographer, between-reader, and between-visit correlations were rP = 0.80 to 0.87, rP = 0.73 to 0.89 and rP = 0.63. The mean absolute value of the difference between replicates was 59.4 to 94.0 cm/sec and 88.7 to 112.8 cm/sec for sonographers and readers, respectively. These results indicate that the mean PWV measure is reproducible even when sonographers and readers are newly trained.