Youden's two sample diagram is a useful method for certain types of interlaboratory comparisons of test results. Generally, points in the plot fall within an elongated ellipse, the major axis of which makes a 45°angle, approximately, with the x, y axes. Occasionally it happens that the axes are not the bisectors of the coordinate axes. This paper (1) examines more closely the assumptions underlying the Youden diagram and presents a more general method of interpreting it and (2) generalizes the diagram to situations where the two samples do not have the same level and/or the axes of the ellipse definitely do not bisect the coordinate axes.