Noise-equivalent power (NEP) is often used to express the voltage noise at the output of a photoreceiver amplifier in terms of an equivalent optical-signal level. However, avalanche photodiode (APD) receivers of the same NEP may differ substantially in terms of the amplitude distribution of the noise, resulting in substantial variation of the detection threshold required to extinguish false alarms. Thus, NEP alone is not a good measure of the sensitivity of a lidar receiver that employs an APD. The optical signal required to achieve a specified detection probability for a detection threshold that achieves a specified false-alarm rate is shown to be a more reliable characterization of APD photoreceiver performance for lidar applications.