The subject chosen is shown increasing interest, as it is extremely important in the practice of both internal and financial auditors. The point of extracting a sample out of a population of documents or processes is that of saving time and reducing costs, maintaining at the same time, the confidence that examining the entire population would have led to the same results. Reviewing the annual reports of a company not only requires sampling skills but also the ability to interpret results in a scientific, probabilistic way. Although professional judgment is considered essential, a more accurate approach to audit procedures is by using statistical sampling methods. In order to show the dramatic effect of using probabilistic methods of determining the sample size over an audit engagement, and interpreting the results adequately, we have chosen to apply two methods. One is based almost entirely on professional judgment, is based on risk analysis and ignores important aspects such as population size, precision and confidence level. The second method relies on all these statistical and probabilistic indicators and also takes into account the nature of the elements of the population being tested. The result is surprising the sample size determined is more than three times larger than the one estimated through the first, risk based method.