Over the years, there have been a number of cases where users experienced higher noise levels of transformers on-site versus what was measured in the factory acceptance tests. In some of these cases, measurements have shown that a power transformer can have significantly different noise characteristics (both noise level and frequency spectrum) on-site compared to what was measured in the factory. This difference can range from a few decibels to more than 10 dB in some cases. Some of these differences are found to be caused by local environmental factors that influenced the on-site measurements. Others were caused by operating conditions on-site that are different from those of the factory tests. This paper presents results of investigations and measurements performed, over the past several years, on the main contributors to this difference. For each of these contributors, data are presented that demonstrate the impact of this contributor as well as quantify typical magnitudes of this impact. It is hoped that the material presented in this paper will assist users in: 1) understanding some of the pitfalls of on-site measurement of transformer noise; 2) how operating conditions contribute to the noise performance of transformers on-site; and 3) how significantly higher on-site noise levels can, in some cases, indicate an issue with the transformer or any of its components/accessories. It is also hoped that the material in this paper will contribute to industry standards/application guides on noise determination and abatement.