However, the lack of significance of BP variability between treated and control groups could reflect the absence of any real difference, or lack of power to show a difference. Makes it probable that thiazides do not decrease BP variability and possible that they could increase BP variability. This should be tested with other databases and preferably databases where individual patient data are available. If there is an increase in variability associated with thiazide treatment it would be important to determine whether it was an increase in interpatient or intrapatient variability. It is particularly the setting of an increase in intrapatient variability that an antihypertensive therapy could increase risk to patients. Systolic and diastolic BP measurements are dependent on each other; however, there are physiologic settings where systolic BP rises more than diastolic such as during exercise and as muscular arteries lose their elasticity as a part of normal aging. In clinical trials,CA3 resting blood pressure is measured in a standardized way. We are not aware of other settings where the variability of systolic and diastolic blood pressure has been directly compared using the coefficient of variation. In this case we have used all the unconfounded estimates of systolic BP and compared them with all the unconfounded estimates of diastolic BP to increase the chance of showing a difference. In the event the systolic CV was statistically significantly greater than the diastolic CV. This may reflect a true physiological difference. However, it is more likely due to an artifact of the method of measurement. In this case measurements were auscultatory using a mercury manometer. This means that the systolic BP is measured first and the diastolic BP is measured after a short delay. Because of this difference in timing of the two measures patient factors could contribute to the difference in variability e.g. Patients are more relaxed as the pressure in the cuff decreases. Alternatively, it is easier to accurately 7-Ethylcamptothecin measure systolic blood pressure than diastolic blood pressure. Thus the difficulty in detecting the disappearance could lead to a greater likelihood of guessing, which would be expected to artificially lower the variability of diastolic blood pressure. It will be important to repeat this analysis with other data and in other settings. For example a study of blood pressures measured with automatic BP machines using an oscillometric technique may not show a difference in variability of systolic and diastolic blood pressure, thus providing evidence in favor of this being an artificial difference caused by the technique of measurement. Whatever the explanation for the statistically greater variability of systolic blood pressure the magnitude of the increase in variability is small and probably not clinically significant. We do not think that it is a reason to suggest that diastolic blood pressure is a more reliable measure. In conclusion, systematic reviews can often reveal much more than the original objective of the work. Blood pressure variability as estimated by SD is an important measure and researchers in the area should be familiar with the average magnitude of that variability, 14 mmHg for systolic and 8 mmHg for diastolic, and the factors that can affect it.