Cultural heritage digitization centers worldwide rely on quality management systems to insure the accuracy and consistency of their digitization efforts. Universal to all digitization quality management is the accurate measurement of color test targets. Testing at the Library of Congress has shown that significant variability exists between manufacturers and models of spectrodensitometers reading standard test targets. This variability has been measured commonly as ΔE2000 3 and as large as ΔE2000 4, leading to improper calibration of imaging systems and less than ideal color accuracy. In this paper we present two robust statistics-based algorithms that estimate the true color from multiple spectrodensitometer measurements, effectively reducing measurement error. The estimated colors can then be regarded as the ground truth for imaging device calibration and performance evaluation. Besides the true color estimation, the second algorithm in particular characterizes the performance of the devices by estimating their bias and variances.