This paper describes an evaluation of an in-the-ear hearing aid, the ReSound model ED2, which applies fast-acting compression independently in two frequency bands. This can compensate for the loudness recruitment typically associated with sensorineural hearing loss. The crossover frequency between the two bands, and the gain and compression ratio in each band are programmable to suit the individual patient. Speech reception thresholds (SRTs) in 12-talker babble were measured for six subjects with moderate sensorineural hearing loss under monaurally and binaurally aided conditions, with the speech and babble both coincident and spatially separated. The ED2 programmed as a two-band compressor gave better speech intelligibility (lower SRTs) than the ED2 programmed as a linear amplifier, and also than unaided listening. A significant advantage for binaural aiding was found, and this advantage was greater when the speech and noise were spatially separated.