Evidence suggests that high peripheral beta-amyloid (Abeta)(1-40) levels and low ratios of Abeta(1-42)/Abeta(1-40) are associated with increased risk for Alzheimer's disease (AD). In this cross-sectional design, serum and plasma samples from 67 AD patients and 146 controls (similar in age and gender) were evaluated using Abeta(1-40) and Abeta(1-42) ELISA. Coefficient of variance was calculated for intra- and inter-person variability of Abeta(1-40) and Abeta(1-42). Abeta(1-40) correlated with age, MMSE and their Abeta(1-42)/Abeta(1-40) ratios (p<0.05). Significantly higher Abeta(1-40) levels were observed in AD patients than controls (p<0.05) but no difference was observed for Abeta(1-42) (p>0.05). Serum Abeta(1-42)/Abeta(1-40) ratios were also significantly lower in AD patients than controls (p<0.05). Lower intra-person than inter-person variability was observed for serum and plasma Abeta(1-40) and Abeta(1-42) and these were higher in controls than in AD patients. The intra-person variability of serum Abeta(1-40) did not influence the group differences observed between AD patients and controls. Significant interaction was observed between diagnosis and intra-person variability for serum Abeta(1-40) levels (p<0.05) and was supported by our finding of higher intra-person variability for serum Abeta(1-40) in controls (26.97%) than in AD patients (18.35%). We confirm the previously observed differences in blood Abeta levels between AD and control groups. In addition, we now report the presence of high intra- and inter-person variability possibly due to factors that influence peripheral Abeta levels and warrant further investigation before the potential use of Abeta as an AD biomarker can be fully exploited.