Objective: To formulate a reliable classification of tympanic membrane retraction which is easy to use but capable of detecting small differences in retraction. Study Design: Prospective study. Methods: The classification was developed from observations dividing the drum into 3 areas: the pars tensa anterior to the malleus, type I, posterior to the malleus (subdivided into upper and lower) type II, and attic type III. The subclassification on paper is more complicated but using a database it is easy to use. The classification is automatically calculated by the database which can also be converted to numerical form. The classification also allows documentation of active disease in retractions. Photographs of seven retracted ear drums were incorporated into the database and ten otolaryngologists asked to classify them. Five were asked to reclassify the retractions in the same way after 3 months. Results: Intra class correlation was significantly high (>0.9) for pars tensa, attic, and for bony erosion. Cronbach's alpha values were also high (>0.9) in all groups. Retest values were evaluated with Wilcoxon's signed rank sum test establishing that there was no significant difference in results. Conclusion: The classification shows reliabiliy and validity allowing detection of small changes in tympanic membrane retraction especially affecting the pars tensa but allows classification of the whole tympanic membrane.