Purpose. Group physical therapy in patients with ankylosing spondylitis was studied to determine whether beneficial effects persisted after cessation of the intervention. Methods. After a 9-month period of supervised group physical therapy, 68 patients were randomized for another 9 months to unsupervised daily exercises at home (discontinuation group) or continuation of weekly sessions of supervised group physical therapy (continuation group). Endpoints were spinal mobility (thoracolumbar flexion and extension, chest expansion, cervical rotation), fitness (maximum work capacity), functioning (Sickness Impact Profile (SIP), Health Assessment Questionnaire for the Spondylarthropathies (HAQ-Sl, Functional Index [FI]), and patient's global health assessment on a visual analogue scale. Results. Time for exercises at home was significantly higher in the continuation than in the discontinuation group (mean duration 1.9 versus 1.2 hr per week, P < 0.05). The continuation group improved in global health (mean improvement 1.6; 32%) and in SIP score. Scores for thoraco-lumbar mobility and HAQ-S did not change very much, whereas chest expansion, cervical rotation, fitness, and FI deteriorated. The average attendance for group therapy sessions was 62%. The discontinuation group improved only marginally (0.2; 4%) in global health, whereas all other endpoints decreased. Only for global health and HAQ-S were the differences statistically significant in favor of the continuation group. Conclusions. Global health and functioning are sustained or even improved further if group physical therapy is continued. Spinal mobility decreased slightly in both groups.