Ten patients with respiratory failure and nocturnal hypoventilation were treated for three to nine months by nasal intermittent positive pressure ventilation. Four patients had chronic obstructive lung disease (median FEV1 19% predicted) and six restrictive chest wall disorders (median FVC 25% predicted); eight of the patients also had cardiac failure. The median daytime arterial oxygen tension, measured before and after at least three months' treatment, increased from 6.2 (range 5.4-9.6) to 9.1 (7.1-9.8) kPa in those with restrictive disease (p less than 0.05), and from 6.0 (5.7-6.5) to 7.1 (6.3-7.7) kPa in the four with airflow limitation (NS). Median values for arterial carbon dioxide tension over the same time fell from 8.2 (range 6.7-9.8) to 6.5 (6.0-6.9) kPa in the group with restrictive disease (p less than 0.05) and from 8.2 (7.0-9.2) to 7.1 (4.9-7.7) kPa in those with airflow limitation (p less than 0.02). Total sleep time while patients were using nasal positive pressure ventilation varied from 155 to 379 (median 341) minutes, and included 4-26% rapid eye movement sleep (median 14%). The percentage of monitored time during the night in which the arterial oxygen saturation was less than 80% fell from a median (range) of 96 (3-100) to 4 (0-9) in the six patients with restrictive disease and from 100 (98-100) to 40 (2-51) in those with airflow limitation. There were no changes in spirometric values but exercise tolerance improved in all patients. The technique may prove an acceptable alternative to long term domiciliary oxygen therapy in selected patients.