Various strategies for representing speech information with multichannel cochlear prostheses were compared in tests with implant patients. The strategies included the compressed analog (CA) approach of a standard clinical device, and alternative interleaved pulses (IP) and continuous interleaved sampling (CIS) strategies. CA and IP strategies had been compared in previous studies with a wide range of subjects. The present studies compared all three types in tests with one subject and CA and CIS strategies in tests with six additional subjects. Subjects for the present studies were selected for their excellent performance with the clinical CA processor, and the tests included closed-set identification of consonants and open-set recognition of words and sentences. For every test, every subject obtained his or her highest score, or repeated a score of 100% correct, using a CIS strategy. In the comparisons of all three approaches, IP processor scores were between those obtained with CA and CIS processors. The results are discussed in terms of their implications for processor design.