Fourteen healthy elderly men had polysomnography performed on two consecutive nights to assess the consistency of sleep and breathing from night to night. The reported first-night effect was seen on electroencephalographic sleep, leading to lighter or fitful sleep on the first night. Mean values for apneas, hypopneas, and oxygen desaturations did not change from night 1 to night 2. Five of 14 subjects would have changed classification from night to night if a cutoff of five apneas and hypopneas per hour were used to define normality. Because this numerical cutoff may not be valid in the elderly, the clinical significance of such a changing classification is not clear.