Reliability and reproducibility of an automatic ambulatory blood pressure recorder (ICR-5200, Squibb) were tested on 45 subjects (24 normotensives and 21 hypertensives) aged 32-64 (mean age 49). Of 6,566 recordings, 2,376 measurements were cross-checked with a Riva Rocci sphygmomanometer in supine, standing, or sitting position at the beginning and at the end of a 24-hour monitoring period. Moreover, in 11 normotensives and in 11 hypertensives, cross-checking was performed during a cycloergometer effort test. A highly significant linear correlation was found in all the conditions tested (r = 0.98, p less than 0.001 for systolic and r = 0.98, p less than 0.001 for diastolic pressure) either in hypertensives or in normotensives. Only 8% of diastolic measurements performed by the two methods differed more than 5 mmHg. For each subject 5.4 +/- 3.8% of the preprogrammed measurements were lost. Only seven out of 45 subjects referred a disturbed sleep. ICR-5200 seems a suitable and reliable tool for automatic blood-pressure recording.