A method for prescribing a nonlinear hearing aid using a simplified loudness model and the speech intelligibility index (SII) model [ANSI draft (1993)] is described. The specific loudness of a signal is calculated as a linear function of the sound-pressure level bounded between threshold and UCL. An integrated-SII model is constructed by computing the weighted sum of the SII values across several listening conditions [Hou and Thornton, Ear Hear. 15, 105--112 (1994)]. The fundamental goal of the method is to adjust the hearing aid parameters so as to maximize the integrated SII. The specific loudness of the aided speech and noise for a hearing-impaired subject is constrained so as not to exceed the specific loudness of the unaided signals for normal-hearing listeners in each listening condition. The method is applied to a database of patients who wear multiband wide dynamic range compression hearing devices. Resulting hearing aid program parameters are compared with the patients' existing programs.