The objective of this study was to determine stability of the neck muscle activation pattern in a given dystonic head position in patients with cervical dystonia (CD). We assessed 26 patients with CD and botulinum toxin (BT) treatment failure before surgical denervation. None of them had received BT injections for at least 4 months. To relate dystonic head position to underlying neck muscle activity, we used synchronised video and poly-electromyographic (EMG) recording over a period of 10 minutes. The muscle activation pattern during constant (“stable”) maximal dystonic excursions was analysed. EMG data of nine patients was excluded from the analysis, as these patients had a constantly changing head position or marked head tremor. In the remaining 17 patients, who had a fairly stable dystonic position, muscular activation patterns during the recording spontaneously changed in nine (Group A) while in eight it remained stable (Group B). There was no significant difference in demographic variables between the two groups other than a male predominance in Group A. However, the retrospectively determined initial response to BT treatment (before BT treatment failure had occurred) was significantly worse in Group A as compared with Group B. Neck muscle activation patterns can spontaneously change in CD patients despite constant dystonic head position, implying an inherent variability of the underlying central motor program in some patients. This should be considered when BT treatment response is unsatisfactory, and should also be taken into account when interpreting results of EMG recordings of neck muscles in these patients. © 2001 Movement Disorder Society.