The aim of the study described here was to evaluate the standard deviation (SD) as a measure of blood-glucose variability in IDDM patients under 'normal life' conditions. One hundred IDDM patients performed self-monitoring of blood glucose (SMBG) five times every 2 days for 4 weeks. From these records the following measurements were calculated for each patient: the standard deviation of all blood-glucose values (SDBG), the M-value, the percentage of values < 3 and > 15 mmol l-1 (PE), and the mean, absolute difference of consecutive blood-glucose values (MAD), a novel measure of blood-glucose variability, also taking into consideration the succession of the values. Before the study the patients as well as their physicians were asked to estimate the blood-glucose stability of the patient, using a five-category scale of statements. The patients recorded an average of 64 (range: 32-70) SMBG values. The SDBG was normally distributed with a mean of 3.9 +/- 1.0 mmol l-1. There was a highly significant correlation between the SDBG and the other measures of blood-glucose variability (p = 0.0001, r > 0.8). It appeared that the variation of the SMBG values recorded before dinner contributed to the total glucose variability to a great extent. There was a poor agreement between the subjective estimations of the blood-glucose stability made by the patients and the physicians and the objective measures of the blood-glucose variability. It is concluded that the SD provides an accurate and easily available estimate of blood-glucose variability in IDDM patients practising home blood-glucose monitoring.