Fish swimming energetics are often measured in laboratory environments which attempt to minimize turbulence, though turbulent flows are common in the natural environment. To test whether the swimming energetics and kinematics of shiner perch, Cymatogaster aggregata (a labriform swimmer), were affected by turbulence, two flow conditions were constructed in a swim-tunnel respirometer. A low-turbulence flow was created using a common swim-tunnel respirometry setup with a flow straightener and fine-mesh grid to minimize velocity fluctuations. A high-turbulence flow condition was created by allowing large velocity fluctuations to persist without a flow straightener or fine grid. The two conditions were tested with particle image velocimetry to confirm significantly different turbulence properties throughout a range of mean flow speeds. Oxygen consumption rate of the swimming fish increased with swimming speed and pectoral fin beat frequency in both flow conditions. Higher turbulence also caused a greater positional variability in swimming individuals (versus low-turbulence flow) at medium and high speeds. Surprisingly, fish used less oxygen in high-turbulence compared with low-turbulence flow at medium and high swimming speeds. Simultaneous measurements of swimming kinematics indicated that these reductions in oxygen consumption could not be explained by specific known flow-adaptive behaviours such as Kármán gaiting or entraining. Therefore, fish in high-turbulence flow may take advantage of the high variability in turbulent energy through time. These results suggest that swimming behaviour and energetics measured in the lab in straightened flow, typical of standard swimming respirometers, might differ from that of more turbulent, semi-natural flow conditions.