The covariation between thigh, shank and foot elevation angles during locomotion was analysed by means of orthogonal planar regression in a patient with pure hereditary spastic paraparesis before and after an intrathecal bolus of baclofen and in seven healthy subjects. The size, shape and spatial orientation of the loop de®ning patient's planar covariation (thigh angle vs. shank angle vs. foot angle) signi®cantly differed from the controls' before baclofen, whereas these features resumed normal characteristics after baclofen injection. This shows that alteration of the control of phase coupling for the co-ordination of lower limb segments in human gait by increased spinal re¯exes can be reversed by intrathecal baclofen injection. q 2000 Elsevier Science Ireland Ltd. All rights reserved.Keywords: Human; Locomotion; Co-ordination; Planar covariation; Baclofen; Hereditary spastic paraparesis As a GABA B agonist [14], baclofen reduces the release by primary afferent terminals in laminae II and III of excitatory neurotransmitters onto ventral horn motoneurons in the spinal cord [5,17]. Although intrathecal baclofen (ITB) is becoming a standard treatment of spinal origin spasticity [15], its effect on locomotor control is unclear. A recent approach has revealed a speci®c covariation of elevation angles of the lower limb segments along an attractor plane during locomotion in healthy humans [2±4,10]. The plane orientation and the shape of the loop that de®nes it re¯ect the phase relationships between these angles and therefore intersegmental co-ordination, on which postural stability with respect to gravity and dynamic equilibrium for forward progression depend. Recently, the features of this covariation in Parkinson's disease before and after therapeutic intervention gave insights into basal ganglia function [11]. In this study, we analysed this covariation in a patient with uncomplicated autosomal dominant hereditary spastic paraparesis (HSP) [9] before and after an ITB bolus.The studied patient, aged 41, has normal muscle power, increased tone and re¯exes in the lower limbs, extensor plantar responses and a spastic gait. Seven healthy subjects (aged 38.2^4.6) participated as controls.Using the ELITE system [8], four sessions of ten trials of the patient's self-paced locomotion over ten meters were recorded with a 100 Hz sampling rate, respectively before ITB of 75 mg (0.77 mg/kg) via lumbar puncture and 2, 4 and 6 h after it. Ten trials were recorded for each control subject. Markers over the anterior-superior iliac spine, trochanter, lateral knee condyle, lateral malleolus and 5th metatarsal, de®ned the segments of the thigh, shank and foot.Statistical analysis of the angle covariation was based on principal component (PC) analysis (see [2]). PCs were computed by pooling the sample of time-varying angles after subtracting the mean. PCs are linear combinations of variates which are the covariance matrix eigenvectors. The ith PC is given by: PCi u i T a where u i is the eigenvector and a the variates. The norm...