The article deals with the concept of within-item variation, given by Arnold Barkman, in both sampling and non-sampling audit situations. It argues that results, which are similar to and perhaps, more appropriate than Barkman's heuristic formulation for within-item variation have been developed previously, the consequence of ignoring correlation among the individual account items is likely to be underestimation of within-tem variation and use of suggested technique for estimation of parameters seems questionable, and perhaps unnecessary. Barkman has incorrectly defined the relationship between the true account value and a current measurement of that account value. It is more proper to consider the true value as a fixed constant and the current measurement as a random variable equal to the sum of the true value and a random error term. The true value may be difficult or impossible to measure with certainty when the audit is performed. An account receivable balance will be fully or partially collected, or fully or partially written-off at some future date. An inventory item will or will not have a market value lower than its cost when it is sold. It is, therefore, appropriate to consider this current period measurement as the value of a random variable whose variability depends on factors such as economic conditions, type of account, and auditor experience and judgment.