Diurnal blood pressure (BP) variation can be assessed by cusums-derived measures and by the day-night BP difference from time-, activity- and diary-defined 'night' periods. Reproducibility of diurnal systolic BP (SBP) variation by these different methods was studied in 19 active elderly normotensives, mean age 68.5 years. Subjects underwent simultaneous 24 h BP (Spacelabs 90207) and activity (Gaehwiler wrist actigraph) monitoring on two occasions (median interval 70 days). On the first occasion, mean diurnal SBP variation was 15.1 +/- 8.1 mm Hg by fixed-time definition of 'night' (22.07). When compared with 22-07 defined 'night' period, actigraph- and diary-defined 'night-time' was significantly reduced (-60 +/- 49, and -48 +/- 51 min, respectively) and consequently diurnal SBP variation was significantly greater at 18.2 +/- 8.1 mm Hg and 17.6 +/- 8.4 mm Hg, respectively. Actigraph recordings were also used to exclude 'night' BP readings associated with activity, but this did not significantly alter the diurnal SBP variation. Cusums-derived circadian alteration magnitude resulted in the greatest value for SBP variation (23.4 +/- 6.7 mm Hg). However, reproducibility of diurnal SBP variation was poor by fixed-time method with a coefficient of variation (CV) of > 50%, and only improved to 40% with diary use. Actigraph measurements, even if used to exclude BP values associated with disturbed sleep, did not improve this further. Cusums-derived measures of diurnal variation slightly improved reproducibility with a CV of 34.6% and may be a better method in the assessment of diurnal BP variation in the elderly.