Objective: Tympanometry and acoustic reflectometry are suggested tools for confirmation of otoscopic diagnosis of secretory otitis media. The issues on sensitivity and specificity of both devices are contradictory. In this study, our purpose was to compare sensitivity and specificity of both devices and to look for whether it is possible to reach higher values by combining them. Methods: This study included 150 normal ears and 150 ears with chronic effusion. In tympanometry, only B tracings were accepted as predictor of effusion. In acoustic reflectometry, reflectivity (cut point: 5) and curve angle with two cut-points (75 degrees and 90 degrees) were used. Results: Acoustic reflectometry presented higher specificity by both reflectivity (cut point: 5) and by curve angle (cut point: 75 degrees) (99.33% by both) than tympanometry (92%) (chi(2) analysis, P < 0.001). But, their sensitivities (65.33 and 78%) were lower than tympanometry (96%) (chi(2) analysis, P < 0.001). With curve angle of 90 degrees, specificity of acoustic reflectometry decreased to 85.33%, sensitivity increasing to 97.33%, which was not different from tympanometry (chi(2) analysis, P > 0.1). When data of curve angle and tympanometry were combined, specificity and sensitivity of the combined test were found to be 91.33 and 100%, respectively. Conclusions: (i) Acoustic reflectometry should not be proposed as a better device than tympanometry, because its test efficiency was not higher than tympanometry. (ii) The only advantage of AR (reflectivity greater than or equal to 5 and curve angle less than or equal to 75 degrees) was its high specificity to effusion. In addition, higher curve angles than 90 degrees were found to be highly predictive for normal ears as much as tympanometry. But. predictivity of curve angle between 76 degrees and 90 degrees was low. (iii) When tympanograms and curve angle were combined, it was found that prediction of this combination for curve angles between 76 degrees and 90 degrees was perfect. (iv) We consider that both test devices provide complementary data to each other, which would be particularly important for screening studies and that they are good tools for confirmation of clinical impression, particularly for less experienced clinicians. (C) 1999 Elsevier Science Ireland Ltd. All rights reserved.