The goal of this dissertation was to understand how the intrinsic dynamics of gait adapt to support the performance of an ecologically relevant object transport task. A common object transport task is walking with a cup of water. Because the water can move relatively independent of the cup, the cup and water system is classified as a complex object. To model this task participants carried a cup with a wooden lid placed on top. On the lid there was a circular region with the same circumference as the cup and a ball. The object of the task was to keep the ball inside the circular region. We explored two questions: 1) how do the intrinsic coordinative gait dynamics adapt to support object transport during walking? And 2) how do individuals adapt to manually control a complex object when asked to concurrently attend to visual information? To address question 1, participants walked on a treadmill at six speeds (0.4 - 1.4 m/s) and performed three conditions: normal walking, walking with a cup only (Cup), and walking with the cup and ball (Cup-Ball). When performing the Cup-Ball condition, as gait speed increased, pelvis-thorax coordination was more in-phase compared to the normal walking and the Cup conditions. Arm-leg coordination was affected by the performance of the Cup-Ball condition. On the constrained side arm-leg coordination was 2:1 while a 1:1 relationship was maintained on the unconstrained side. A correlation between the amplitude of the unconstrained arm and manual task performance revealed a significant negative correlation as gait speed increased, indicating that individuals who reduced their arm swing performed better. To address question 2, participants walked on a treadmill at three gait speeds under four task conditions: normal walking, walking with the cup and ball system (Cup-Ball), walking while identifying visual stimuli (Visual), and a combined condition where participants walked with the cup and ball system while identifying visual stimuli (Cup-Ball-Vis). The addition of the visual task in study 2 resulted in the head orientation to be more extended relative to the trunk with a larger range of motion compared to the manual task only condition; participants optimized on the visual task at the expense of manual task performance. In both manual task conditions pelvis-thorax coordination was more in-phase as gait speed increased and more variable compared to the walking only condition. The latter result demonstrates the functionality of increased coordination variability during object transport tasks. The amplitude of the unconstrained arm decreased as the system became more constrained (i.e., going from walking only to Cup-Ball to Cup-Ball-Vis tasks). Although the arm amplitude decreased, the unconstrained arm maintained a 1:1 arm-leg coordination while the constrained arm was in a 2:1 relationship for both manual task conditions. This result demonstrates that the unconstrained arm continues to move to counteract angular momentum imparted by the legs while the arm carrying the object is coupled to the step frequency, counteracting disturbances imposed by heel contacts. The overall results from both studies demonstrate that the body’s natural walking dynamics adapt to support manual task performance. The segments not directly involved in the task continue to interact to maintain intrinsic gait dynamics. This dissertation makes significant contributions to the literature by demonstrating: 1) asymmetries in arm-leg coordination are exploited by the body to maintain manual task performance and intrinsic gait dynamics; 2) amplitude of the freely swinging arm is an important factor in task performance during object transport; and 3) increased variability at the level of the pelvis-thorax interaction plays a functional role in maintaining both manual and visual task performance. The significance of the findings here is that they demonstrate how task constraints alter intrinsic coordination dynamics during walking in order to support performance while at the same time maintaining gait stability