There is a lack of reliable and easy-to-use methods for the quantitative determination of carbohydrates in natural waters. Among the existing methods, the 3-methyl-2-benzothiazolinone hydrochloride (MBTH) method has been widely used to determine total dissolved carbohydrates in seawater. Its application to freshwaters has, so far, been less frequent. The objective of this study is to critically examine the application of the MBTH method to the analysis of carbohydrates in freshwater samples in order to understand what the method really measures. Following a comprehensive revision of the literature, the method has first been applied to representative model compounds and then to a variety of freshwaters of contrasting trophic characteristics. By simultaneously determining the total dissolved organic carbon and humic-type compounds in the same samples, it has been possible to show that sometimes a significant part of the organic carbon remains undetected. This seems to indicate that a substantial amount of carbohydrate present in some natural waters is probably not ‘seen' by the MBTH method.