Objective Purpose of this study was to find out test accuracy and predictivity of tympanometry (TYMP) and parameters (reflectivity and curve angle) of acoustic reflectometry (AR) in the ears with prior clinic and otoscopic evidences of effusion in which no effusion was detected by myringotomy, by comparing with the data found in normal ears and in those with effusion. Methods: While study group included 31 ears with chronic retraction in which no effusion was detected by myringotomy, control group included 150 fully normal ears and 150 ears with effusion confirmed by myringotomy. B tracings in TYMP, and values of reflectivity higher than 5 and curve angle lower than 75 degrees in AR were accepted as indicators of effusion; in the combined test, the ears with non-B tracings in which curve angle higher than 75 degrees were accepted as those without effusion. Results: False positivity values of TYMP, reflectivity and curve angle and the combined test found in the ears with chronic retraction without effusion (29, 23, 19 and 35% respectively; chi(2)-test, P > 0.1) were significantly higher than normal ears (chi(2)-test, P < 0.001), but lower than those of effusion (chi(2)-test, P < 0.0001). Further, although negative predictivity values of TYMP, reflectivity and curve angle were found to be lower in these ear, compared to normal ears (chi(2)-test, P < 0.01), the combined test reached to a perfect negative predictivity, as much as in normal ears. On the other hand, positive predictivity values were decreased in the ears with chronic retraction without effusion by only reflectivity. Conclusions: By both TYMP and AR alone, it is more difficult to find out the ears without effusion among the ears clinically and otoscopically diagnosed as otitis media than to find out normal ears. Therefore, in clinical work, false positivity of these devices were higher, and their specificity and negative predictivity were lower; although the combined test solved problem of low negative predictivity, high number of false positive ears in clinical work, that is lower specificity, continues to be an unsolved problem by either TYMP or AR alone, or both together. (C) 2000 Elsevier Science Ireland Ltd. All rights reserved.