The treatment of positional plagiocephaly is controversial. A confounding factor is the lack of a proven clinically viable measure to quantify severity and change in plagiocephaly. The use of anthropometric measurements is one proposed method. In this study, the reliability and validity for this method of measurement were investigated. Two clinicians independently recorded caliper measurements of cranial vault asymmetry (CVA) for infants referred for plagiocephaly or torticollis, and an unbiased observer recorded visual analysis scores during the same visit. CVA scores were assigned into three predetermined severity categories (normal CVA < 3 mm, mild/moderate CVA <or= 12 mm, moderate/severe CVA > 12 mm). CVA measurements and visual analysis scores were recorded for 71 and 54 infants, respectively. Intrarater reliability was established (kappa = 0.98, kappa = 0.99), but inter-rater reliability was not (kappa = 0.42). In addition, the inter-rater reliability for the severity categories based upon these measures was poor (kappa = 0.28) and failed to correlate to the visual analysis (kappa = 0.31). Development of a stable and meaningful measurement system for the extent of plagiocephaly is needed to allow scientific studies of the natural history of plagiocephaly and effectiveness of interventions.