A portable sound processor has been developed at the University of Melbourne for use with the 22-electrode cochlear implant manufactured by Cochlear Pty Limited. In this report, the performance of the Spectral Maxima Sound Processor (SMSP) is compared with the Mini Speech Processor (MSP) presently used clinically with this implant. Results of tests with 5 subjects demonstrate that the performance of the SMSP significantly exceeds that of the MSP for recognizing vowels and consonants in closed-set confusion studies and open-set monosyllabic word tests, although recognition of speaker identity and intonation patterns is similar for both processors. Information transmission analysis of phoneme features confirms that most features, such as vowel format frequencies and consonant manner and place of articulation, are better conveyed with the SMSP. Possible reasons for its improved performance are discussed.