To evaluate the safety and effectiveness of subsymptom threshold exercise training for the treatment of post-concussion syndrome (PCS). Prospective case series. University Sports Medicine Concussion Clinic. Twelve refractory patients with PCS (6 athletes and 6 nonathletes). Treadmill test to symptom exacerbation threshold (ST) before and after 2 to 3 weeks of baseline. Subjects then exercised 5 to 6 days per week at 80% ST heart rate (HR) until voluntary peak exertion without symptom exacerbation. Treadmill testing was repeated every 3 weeks. Adverse reactions to exercise, PCS symptoms, HR, systolic blood pressure (SBP), achievement of maximal exertion, and return to work/sport. Pretreatment, ST occurred at low exercise HR (147 + or - 27 bpm) and SBP (142 + or - 6 mm Hg). After treatment, subjects exercised longer (9.75 + or - 6.38 minutes to 18.67 + or - 2.53 minutes, P = .001) and achieved peak HR (179 + or - 17 bpm) and SBP (156 + or - 13 mm Hg), both P < .001 versus pretreatment, without symptom exacerbation. Time series analysis showed significant change in rate of symptom reduction for all subjects and reduced mean symptom number in 8/11. Rate of PCS symptom improvement was related to peak exercise HR (r = -0.55, P = .04). Athletes recovered faster than nonathletes (25 + or - 8.7 vs 74.8 + or - 27.2 days, P = .01). No adverse events were reported. Athletes returned to sport and nonathletes to work. Treatment with controlled exercise is a safe program that appears to improve PCS symptoms when compared with a no-treatment baseline. A randomized controlled study is warranted.