Carload waybill statistics as collected by the Interstate Commerce Commission and the Federal Railroad Administration have been the primary data source on railroad commodity flows for more than three decades, and a considerable amount of geographic research during that time period has made use of these data. Using recently released carload waybill data at the freight station level and actual flow data for the Penn Central, Norfolk and western, and Southern railroads, a statistical comparison of the two data sets is undertaken. On the basis of correlation coefficients the waybill sample does a good job of estimating the universe. However, an examination of the standard errors indicates values large enough to cause some concern.

It has been 30 years since Ullman (.! _) introduced geographers and transport planners and analysts to a set of state-to-state railroad flow data referred to as the 1 percent Carload Waybill Sample. He used the sample as his major data source in the classic American Commodity Flow (~ l, and in later years he was involved in efforts to improve the quality· of these data (1_).