Background: Patients with medically unexplained symptoms (MUS) are frequent in primary care and substantially impaired in their quality of life (QoL). Specific training of general practitioners (GPs) alone did not demonstrate sustained improvement at later follow-up in current reviews. We evaluated a collaborative group intervention. Methods: We conducted a cluster randomized controlled trial. Thirty-five GPs recruited 304 MUS patients (intervention group: 170; control group: 134). All GPs were trained in diagnosis and management of MUS (control condition). Eighteen randomly selected intervention GPs participated in training for a specific collaborative group intervention. They conducted 10 weekly group sessions and 2 booster meetings in their practices, together with a psychosomatic specialist. Six and 12 months after baseline, QoL was assessed with the Short-Form 36. The primary outcome was the physical composite score (PCS), and the secondary outcome was the mental composite score (MCS). Results: At 12 months, intention-to-treat analyses showed a significant between-group effect for the MCS (p = 0.023) but not for the PCS (p = 0.674). This effect was preceded by a significant reduction of somatic symptom severity (15-item somatic symptom severity scale of the Patient Health Questionnaire, PHQ-15) at 6 months (p = 0.008) that lacked significance at 12 months (p = 0.078). As additional between-group effects at 12 months, per-protocol analyses showed less health anxiety (Whiteley-7; p = 0.038) and less psychosocial distress (PHQ; p = 0.024); GP visits were significantly (p = 0.042) reduced in the intervention group. Conclusions: Compared to pure GP training, collaborative group intervention achieved a progressive, clinically meaningful improvement in mental but not physical QoL. It could bridge gaps between general practice and mental health care.