Four separate acoustic telemetry experiments were conducted during 1979 and 1980 off eastern San Clemente Island in water 1350 to 1450 m deep. In each experiment, telemetry originating near the seafloor was transmitted to receivers aboard a drifting surface ship which remained within 45° from vertically above the telemetry source. Projector and receiver full beamwidths ranged from 40° to 80°. Telemetry formats used were 128 line, 7700 bit/second (b/s) video; 256 line, 11 500 b/s video; 2400 b/s pseudorandom differential quad phase shift keying (DQPSK) and 4800 b/s DQPSK. The first three formats were simultaneously modulated on 8- and 40-kHz carriers while the latter occupied the 8–14-kHz band. Two millisecond pings and cw transmissions were interspersed with the telemetry to provide for acoustic channel monitoring. Video received during nearly vertical transmissions was subjectively free of acoustically induced defects. At angles less than 15°, 1.4 × 107 bits were transmitted error free at 4800 b/s, indicating an underlying error rate ⩽2.14 × 10−7 at the 95% level of confidence. Greater error rates and poorer video, due to beam pattern and power limitations, were observed at larger angles.