A retrospective analysis was undertaken to estimate the effect on measurements of cup-disc ratio (CDR) that result from overall differences among examination methods used by experienced investigators. The findings suggest large inter-observer differences, that is, large differences in measurement attributable to differences in definitions and methods between one observer and another. Extreme caution is indicated before attaching wide clinical or investigative importance to a specific value of CDR. A value of CDR regarded as the dividing line between normalcy and suspicion by one observer may be a highly inappropriate cutoff for another observer. Similarly, if different observers record measurements on the same patient, it may be impossible to determine whether the cup of an individual patient changes with time. Interobserver differences also can cast doubt on conclusions drawn from the comparison of data collected by different investigators. There is a need for standardization of definitions and methods in such clinical or research situations.