Recreational-grade side-scan sonar (SSS) has become an invaluable tool for inland fisheries, particularly when characterizing underwater habitat, but it is being increasingly used for enumerating large-bodied (> 1 m total length [TL]) aquatic fauna. We used SSS in river and reservoir environments to evaluate methods for identifying and counting Paddlefish Polyodon spathula, a large pelagic planktivore of recreational and economic importance that can exceed 2 m in length and weigh over 70 kg. We assessed accuracy and precision among readers to identify Paddlefish by assigning confidence scores (1–3; with 3 being more confident) to sonar images of a ballistics-gel filled fiberglass replica Paddlefish. Readers varied in their confidence scores for the replica Paddlefish and no reader could identify the target beyond 25 m from the transducer. Afterwards, we used SSS to survey several kilometers of a reservoir during summer residency and a large river during springtime spawning migrations. Two readers counted Paddlefish images in the SSS recordings and we estimated population size in the surveyed area with distance sampling. In the reservoir, the number of Paddlefish counted ranged from 172 to 184. In the river, the number of Paddlefish counted ranged from 165 to 617. The exponential model of distance was most-supported for detection in both environments, except there was support for a half-norm distribution for one reader in the river. In the reservoir, abundance estimates were statistically similar between readers at approximately 1500 (7/ha) in the total scanned area. In the river, similar abundance estimates were obtained with the half-norm model from one reader compared to the exponential model of the other reader, resulting in approximately 1500 individuals (30/ha) in the surveyed area. The application of SSS to count Paddlefish has some clear advantages to traditional methods, such as gill netting, and can be done at multiple times of the year. Distance sampling methods compensated for differences in counts among readers, indicating distance sampling can produce similar abundance estimates even when variation in counts exists among readers.