The role and optimal use of audiometry in monitoring for cisplatin ototoxicity are incompletely defined. Audiograms were obtained from 217 patients before treatment with cisplatin-based chemotherapy for cancers of the esophagus, lung, or head and neck. Posttreatment audiometry then was conducted in 53 of these patients. Chemotherapy consisted of two (87%) or three (13%) courses of cisplatin at a dose of 20 mg/m2/day given as a continuous intravenous infusion over 4 days. Simultaneous 5-fluorouracil or paclitaxel also was given, and 38% received concurrent radiation therapy to the head and neck. Air-conduction thresholds for each ear were obtained at 250, 500, 1000, 2000, 4000, 6000, and 8000 Hz. Three three-frequency pure-tone averages (PTA) also were calculated. Framingham gender-specific, age-adjusted norms were used, beginning at age 60 to correct for presbycusis, and the upper limit of normal was calculated as the greater of the Framingham mean plus twice the standard error, or 25 dB. Hearing abnormality was defined as a threshold >10 dB above the norm for any PTA, or >20 dB above the norm for any individual frequency. Hearing loss was defined as an elevation over baseline threshold of >10 dB for any PTA or >20 dB for any individual frequency. Of the 217 patients who underwent baseline testing, 57 (26%) were found to have hearing abnormality in excess of the expected presbycusis. Post-cisplatin audiograms demonstrated hearing loss in 19 of the 53 retested patients (36%) when compared with their own baseline. As determined by tympanometry, none of these subjects had a conductive component to their hearing loss. These observations were independent of the duration of follow-up after treatment and of the total dose of cisplatin administered. The authors conclude that significant preexisting hearing abnormality is common in this patient population and that, even after low-dose cisplatin administration, additional hearing loss occurs frequently. Baseline testing is mandatory if follow-up studies are to be adequately interpreted.