The distinction between the performance of a classifier and the accuracy of the resulting thematic map is discussed. Sampling the map itself is recommended as the preferred method for testing its accuracy. It is shown, nonetheless, that map accuracy can also be determined from the performance of a classifier, as assessed from testing data, provided the prior probabilities of class membership are well estimated. Unfortunately, that is rarely done in practice so that errors are often made in reported map accuracy figures if they are inferred from classifier behavior. Examples are presented to illustrate how an apparently well-behaved classifier can give poor results for certain classes, because of the proportions of those classes in the region being imaged.