Psychophysical detection of electrical stimulation of the cochlea was studied as a function of electrical-current configuration. Subjects were postlingually deaf humans with Nucleus 20 + 2, Nucleus 22, and Ineraid cochlear implants and nonhuman primates unilaterally deafened and implanted with a multielectrode array similar to the Nucleus implant. In nonhuman primate and human Ineraid subjects, which had percutaneous connectors, we compared threshold functions for sinusoids and pulse trains for quadrupolar, bipolar, monopolar, and parallel multipolar stimulation. Thresholds decreased across this set of configurations. In some cases, the effects of current configuration were dependent on sinusoidal frequency and pulse duration. Pulse duration-dependent effects were also seen when comparing bipolar, monopolar, and common-ground configurations. Bipolar and monopolar stimulation were compared in Nucleus subjects using pulse trains at 50 microseconds per phase. For bipolar stimulation, thresholds decreased as a function of electrode separation, reaching a level near that for monopolar stimulation at separations of 3.5 to 6.5 mm in most cases. These results may be interpreted in terms of effects of current configuration on the magnitude and shape of electrical-potential fields produced in the cochlea, although more central factors also play a role in determining psychophysical detection thresholds.