Segment power can be estimated from the rate of change of segment ener
gy or from segment endpoint dynamics. The inequality between these two
calculations is termed the segment power imbalance (Delta P). Explana
tions for this imbalance have been largely anecdotal. We used 3D gait
data for 20 young, healthy adults to show experimentally that Delta P
is always zero when rigid body assumptions are strictly adhered to. We
show mathematically and experimentally that Delta P is non-zero (rang
e: < 1-100 W) when segment endpoints move relative to the segment cent
er of mass (COM). We also show that Lagrangian numerical differentiati
on of segment endpoint trajectories produce small velocity artifacts w
hich result in observable Delta P (range: 0.3-5 W). The benefits of ob
taining a small Delta P was found to be offset by large discrepancies
(range: 8-54 W) in joint power across the joint gap. For the foot, lar
ge Delta P (range: 23-106 W) was observed when the distal velocity com
ponent included velocity of the center of pressure relative to the foo
t's COM. Our results suggest that, unless interpreted carefully, segme
nt power calculations can potentially lead to erroneous clinical judge
ment. (C) 1998 Elsevier Science B.V. All rights reserved.