Errors in stratigraphic measurement may be compensating or non-compensating. Compensating errors do not affect the mean value of the attribute under study, but they do inflate the apparent variability of the observational data. Non-compensating errors affect the apparent mean value, and they may distort facies gradients or trends. The more important kinds of error that may enter facies maps are presented in a tabular summary. The importance of operational definitions in stratigraphic measurement is stressed, and the properties of the resulting numbers are reviewed in terms of the scales of measurement involved. Accuracy and precision, as they apply to stratigraphic data, are touched upon. A pragmatic method for evaluating facies maps as predicting devices is described. This method takes into account all sources of variability in the original map, including not only errors of measurement and judgment, but the natural large scale and small scale stratigraphic variations as well. An example is used to illustrate the expected average magnitude of the uncertainty in predictions from a specific facies map.