Sampling presents an auditor with a possibility to collect evidence and make a conclusion after having tested only a part of population. Having selected a concrete method of sampling, an auditor evaluates whether the activity of an entity under audition, the nature of accounting and audit data approve this method and whether the results would be reliable. If the selected sampling method is unsuitable or applied incorrectly, there is a greater risk that the conclusions made by an auditor can differ if the same audit procedures would be applied to the whole population. The most widely used method of statistical sampling is monetary unit sampling. This method can be applied by every auditor, since no special knowledge or technical means are required. However, in some cases, the results can be inaccurate. The aim of the article is to examine the limitations of the monetary unit sampling method and to suggest a more developed formula of this method. In the process of investigation, it was defined that application possibilities of the monetary unit method are limited by the necessity to determine whether all items in the population are suitable for selection; also, the application of this method is aggravated by an existence of zero or negative values in population. What is more, if an auditor would lack information about the population under audit, we would suggest to apply a correction factor. It allows for evaluating the size of expected misstatement more precisely.