The role of arm swing in running has been minimally described, and the contributions of arm motion to lower extremity joint kinematics and external force generation are unknown. These contributions may have implications in the design of musculoskeletal models for computer simulations of running, since previous models have usually not included articulating arm segments. 3D stance phase lower extremity joint angles and ground reaction forces (GRFs) were determined for seven subjects running normally, and running under two conditions of arm restraint. When arm swing was suppressed, the peak vertical GRF decreased by 10–13% bodyweight, and the peak lateral GRF increased by 4–6% bodyweight. Changes in peak joint angles on the order of 1–5 deg were observed for hip flexion, hip adduction, knee flexion, knee adduction, and ankle abduction. The effect sizes (ES) were small to moderate for most of the peak GRF differences, but large for most of the peak joint angle differences. These changes suggest that suppression of arm swing induces subtle but statistically significant changes in the kinetic and kinematic patterns of running. However, the salient features of the GRFs and the joint angles were present in all conditions, and arm swing did not introduce any major changes in the timing of these data, as indicated by cross correlations. The decision to include arm swing in a computer model will likely need to be made on a case-by-case basis, depending on the design of the study and the accuracy needed to answer the research question.