The accuracy of 3D measurement using a single image of a four-point coplanar target of known size is studied. The factors that influence accuracy are analyzed and experimental techniques are used to establish their individual effect on accuracy. Experiments shows that the four-point coplanar target has two solutions for real images and that mixing these causes large errors. It is concluded that when the target is approximately parallel to the image plane, the accuracy is significantly better compared to other target orientations, and that image distance is a critical factor affecting accuracy in the depth, while the effect of the other factors on accuracy is relatively insignificant. A new calibration technique is introduced that determines an average image distance over the image plane within the depth of focus and improves accuracy at close range.