There is a growing interest in extracting energy from tidal flows using in-stream kinetic energy conversion devices, and among the many questions are the possible effects on marine mammals. Underwater sound is used as a tool for detecting marine mammal presence via their vocalisations, but such Passive Acoustic Monitoring (PAM) requires an understanding of site-specific acoustic detection ranges, and the naturally occurring ambient noise in high-flow environments imposes constraints on detectability. A pilot experiment was carried out at a proposed small-scale (<2 MW) tidal energy site (Grand Passage, Nova Scotia, Canada) to partially assess the feasibility of a local PAM system. One goal was to determine the effective detection range as a function of tidal phase using sounds from a drifting underwater sound projector and moored hydrophones. A co-located acoustic Doppler flowmeter registered the near-bed water velocity. The maximum observed detection range was 700 m, with a false alarm rate of 50%. A second goal was to try different arrangements for passively reducing the effects of flow noise on the hydrophone signal. Ambient noise levels were computed for 4 different frequency bands: 0-2 kHz; 2-20 kHz; 20-50 kHz, and 50-200 kHz. On 0.5 to 1 h time scales, the bottom pod data in the 0-2 kHz band exhibited the highest variability, associated mainly with ferry and other boat traffic, and the 50- 200 kHz band the least. At tidal frequencies, the clearest dependence of noise level on flow speed was in the bare hydrophone data, with the strongest dependence in the highest band, from about 48 dB re 1 μPa at slack water to about 67 dB re 1 μPa at peak tidal flow. At frequencies above 2 kHz, the data from a drifting hydrophone exhibited a pronounced dependence on position along the channel axis, comparable in magnitude to the tidal variation registered by the moored hydrophone.