The Boeing McDonnell Douglas Human Modelling System was evaluated in two experiments. One involving fit of manikins to actaul body scans and conventional data, and the other comparing manikin reach envelopes with human data. Systematic deviations in proportions were found, such as overestimation of sitting height and underestimation of sitting depth and hip width. Reach envelopes were roughly of the correct size, but somewhat displaced, which was traced back to erroneous shift of the shoulder joint. It is concluded that proportions, shoulder joint and lack of compression of soft tissue created undesirable deviations.