Stride durations in gait exhibit long-range correlation (LRC) which tends to disappear with certain movement disorders. The loss of LRC has been hypothesized to result from a reduction of functional degrees of freedom of the neuromuscular apparatus. A consequence of this theory is that environmental constraints such as the ones induced during constant steering may also reduce LRC. Furthermore, obstacles may perturb control of the gait cycle and also reduce LRC. To test these predictions, seven healthy participants walked freely overground in three conditions: unconstrained, constrained (constant steering), and perturbed (frequent 90° turns). Both steering and sharp turning reduced LRC with the latter having a stronger effect. Competing theories explain LRC in gait by positing fractal CPGs or a biomechanical process of kinetic energy reuse. Mediation analysis showed that the effect of the experimental manipulation in the current experiment depends partly on a reduction in walking speed. This supports the biomechanical theory. We also found that the local Hurst exponent did not reflect the frequent changes of heading direction. This suggests that the recovery from the sharp turn perturbation, a kind of relaxation time, takes longer than the four to seven meters between successive turns in the present study.