The improvement in the severity of obstructive sleep-related breathing disorders during nasal continuous positive airway pressure (NCPAP) therapy can account for the decrease in the required NCPAP level with time. The aim of this study was to prospectively quantify the changes in the required NCPAP level over time of use in sleep apnoea-hypopnoea syndrome (SAHS). Forty sleep apnoea-hypopnoea patients were evaluated before and during the time course of NCPAP therapy. The effective NCPAP level was defined as the positive pressure level that abolished apnoeic and hypopnoeic events and snoring in all sleep stages and sleep positions. This pressure level was determined within 2 weeks after baseline diagnostic sleep study. Sleep studies with NCPAP and NCPAP titration were performed after 2 (n = 40), 8 (n = 40), and 20 (n = 24) months of NCPAP therapy. The initial effective NCPAP level was 9.6 +/- 0.4 cmH2O. It progressively decreased to 8.8 +/- 0.4, 7.9 +/- 0.4 and 7.7 +/- 0.5 after 2, 8 and 20 months, respectively; the difference being significant between the first three NCPAP nights. There was a poor relationship between the changes in the effective NCPAP and changes in weight recorded at the different visits. There was a weak negative relationship between the changes in NCPAP and the previous NCPAP level. In 13 patients, the apnoea-hypopnoea index (AHI) remained > 10 n.h-1 at the first NCPAP trial because the effective NCPAP level was not tolerated. Despite a suboptimal NCPAP level, their sleep architecture improved, and they all reported a subjective improvement in diurnal hypersomnolence.(ABSTRACT TRUNCATED AT 250 WORDS)