1. Existing methods of assessing nocturnal episodic hypoxaemia are either insensitive or ignore the majority of the available data.
2. We describe a method of analysis using offline digital processing. A distribution of oxygen saturation (Sao2) with time is produced from all the available data, and subjected to moment analysis to produce a simple index which describes an entire night's Sao2.
3. Our results suggest that the mean and the coefficient of skew fully described a night's Sao2. However, in subjects with chronic air-flow obstruction, the third moment about 100% oxygen saturation (M3100), a single figure, has the same descriptive power as mean and skew.
4. In 17 subjects with chronic air-flow obstruction a significant correlation was found between both daytime Sao2 and Paco2 when plotted against either the M3100 or the skew.
5. Measurements made on two occasions in seven subjects showed good reproducibility for the skew and M3100 indices.