nent of the total variation (CVT) and can be considered in terms of intra-individual (BVI) and inter-individual (BVG) variation. For a given individual the BVI can be determined from the total variation (CVT) and the analytical variation (CVA). Since the CVA can be readily calculated for most analytes the BVI can be readily determined. BVI should re£ect ‘random’ variation; it should be independent of method and analytical variance. Non-random variations such as eiects of diet, menstrual cycle, seasonal variation, age, sex, etc, must be taken into account before the BV is determined. Preanalytical variables, such as collection tube type, sampling technique, etc, also need to be minimized to enable an accurate BVI to be determined. In this edition of the Annals of Clinical Biochemistry, Carmen RicoŁ sa nd colleagues 4 refer to their previously established database which lists BV1 and BVG for many analytes (http://www.westgard.com/biodatabase1.htm). They took care in the compilation of the data base to ensure that the BV were all determined taking into account the non-random and pre-analytical variables and employing the correct statistical analysis on the appropriate numbers of samples and subjects. In this way they have produced a database of mean BVs which can be used to help in the determination of reference ranges and the calculation of what are signi¢cant changes in an analyte, termed the Reference Change Value (RCV), a vital component when monitoring patients. The RCV for any analyte is usually determined on healthy individuals. Carmen RicoŁ sa nd colleagues 4 ask if the RCV was the same in disease as it is in health. If the RCV does change in disease, this will have a fundamental bearing on how we determine what are clinically signi¢cant changes when monitoring patients. From their literature review, we can be reassured that the majority of BVs remain the same in disease as in health. Using PSA as an example, 5 the BV has been shown to be the same in healthy men and in those with stable prostate cancer. Employing the BVI in healthy men along with average CVA for the majority of PSA assays commercially available, a change of PSA of 50% or greater can be considered clinically signi¢cant. This ¢gure is commonly used by laboratories and clinicians. However, RicoŁ s et al. do provide evidence that nine analytes have diierent BV in health and disease. These are: