METHODS Prospective clinical study of two treatments. OBJECTIVE To compare mechanical ventilation (MV) with phrenic nerve stimulation (PNS) for treatment of respiratory device-dependent (RDD) spinal cord-injured (SCI) patients. METHODS Department for spinal cord-injured patients of an insurance-company-run trauma hospital in Hamburg, Germany. METHODS Prospective data collection of treatment-related data over 20 years. RESULTS In total, 64 SCI-RDD patients were treated during the study period. Of these, 32 of the patients with functioning phrenic nerves and diaphragm muscles were treated with PNS and 32 patients with destroyed phrenic nerves were mechanically ventilated. Incidence of respiratory infections (RIs per 100 days) prior to use of final respiratory device was equal in both groups, that is (median (interquartile range)) 1.43 (0.05-3.92) with PNS and 1.33 (0.89-2.21) with MV (P=0.888); with final device in our institution it was 0 (0-0.92) with PNS and 2.07 (1.49-4.19) with MV (P<0.001); at final location it was 0 (0-0.02) with PNS and 0.14 (0-0.31) with MV (P<0.001). Thus, compared to MV, respiratory treatment with PNS significantly reduces frequency of RI. Quality of speech is significantly better with PNS. Nine patients with PNS, but only two with MV, were employed or learned after rehabilitation (P=0.093). The primary investment in the respiratory device is higher with PNS, but it can be paid off in our setting within 1 year because of the reduced amount of single use equipment, easier nursing and fewer RIs compared to MV. CONCLUSIONS PNS instead of MV for treatment of SCI-RDD reduces RIs, running costs of respiratory treatment and obviously improves patients' quality of life.