The biomechanical and neuromuscular mechanisms used by different animals to generate turns in flight are highly variable. Body size and body plan exert some influence, e.g. birds typically roll their body to orient forces generated by the wings whereas insects are capable of turning via left-right wingbeat asymmetries. Turns are also relatively brief and have low repeatability, with almost every wingbeat serving a different function throughout the change in heading. Here we present an analysis of Anna's hummingbirds (Calypte anna) as they fed continuously from an artificial feeder revolving around the outside of the animal. This setup allowed for examination of sustained changes in yaw without requiring any corresponding changes in pitch, roll or body position. Hummingbirds sustained yaw turns by expanding the wing stroke amplitude of the outer wing during the downstroke and by altering the deviation of the wingtip path during both downstroke and upstroke. The latter led to a shift in the inner-outer stroke plane angle during the upstroke and shifts in the elevation of the stroke plane and in the deviation of the wingtip path during both strokes. These features are generally more similar to how insects, as opposed to birds, turn. However, time series analysis also revealed considerable stroke-to-stroke variation. Changes in the stroke amplitude and the wingtip velocity were highly cross-correlated, as were changes in the stroke deviation and the elevation of the stroke plane. As was the case for wingbeat kinematics, electromyogram recordings from pectoral and wing muscles were highly variable, but no correlations were found between these two features of motor control. The high variability of both kinematic and muscle activation features indicates a high level of wingbeat-to-wingbeat adjustments during sustained yaw. The activation timing of the muscles was more repeatable than the activation intensity, which suggests that the former may be constrained by harmonic motion and that the latter may play a large role in kinematic adjustments. Comparing the revolution frequency of the feeder with measurements of free flight yaws reveals that feeder tracking, even at one revolution every 2 s, is well below the maximum yaw capacity of the hummingbirds.