To determine the day-to-day intraindividual variability of fasting plasma glucose (FPG) in newly diagnosed Caucasian type 2 diabetic subjects. A total of 193 newly diagnosed, previously untreated, Caucasian type 2 diabetic subjects (135 men, 58 women) had FPG measured on two consecutive days (FPG1, FPG2). Ethical approval and subjects' full informed consent were obtained. Subjects fasted for 12 h before each study day and rested for at least 30 min before blood was taken. Plasma glucose was analyzed by a glucose oxidase method with intra- and interassay coefficients of variation (CVs) < 2%. Variability of FPG was assessed by comparison of percentage differences (PDs): PD = 100 (FPG2 - FPG1)/FPG1, with averaged FPG (FPGaver = [FPG1 + FPG2]/2). Biological and analytical variability were determined by use of SD2total = SD2biological + SD2analytical, where SD2analytical approximately equal to 2 x (CVglucose measurement)2. Given normally distributed data with zero mean, 95% of daily percentage differences will be expected to fall within a range of +/- 2 SDtotal. Subjects were age 54 +/- 10 years (mean +/- SD) and had BMI of 29.3 +/- 5.3 kg/m2. FPG values for both days were 12.2 +/- 3.4 mmol/l (FPG1) and 12.1 +/- 3.3 mmol/l (FPG2), with a mean paired difference (95% CI) of 0.1 (0.0 to 0.3) mmol/l. The variance of these differences increased with increasing FPGaver. The PDs did not exhibit this effect and were normally distributed (mean -0.6% [-1.7 to 0.4]; SD 7.4% [6.8 to 8.3]), giving a 95% variability (2 SD) of 14.8%. Biological variability (2 SDbiological) was 13.7%. No significant difference in PD was found between men and women (mean difference 1.3% [-1.0 to 3.6]; SDmale 7.4%, SDfemale 7.3%; P = 0.62). A total of 95% of the FPG values for this group of newly diagnosed type 2 diabetic subjects varied within approximately +/- 15% on a daily basis, with approximately 14% caused by biological variability. As these results are expressed in percentage terms, subjects in the group with higher FPG values are likely to experience larger changes in FPG values measured from day to day. This variability should be considered when using FPG for the diagnosis and/or monitoring of response to treatment in patients with type 2 diabetes.