ABSTRACT The movement and spreading of the oil from the tanker Exxon Valdez accident was a vital issue in the massive response activities launched by industry, private groups, state governments and the federal government. To address this issue, a number of observational and forecast programs were carried out. These accumulated an unprecedented amount of observational data ranging from visual and remote-sensing distribution information to fairly complex adaptations of standard trajectory analysis routines. This paper reports on a hindcast study that used observational data and computational procedures to provide a quantitative estimate of the oil distribution over time. The major difficulty in doing this work was associated with the fact that both the observational and computational data had errors. Typically, the observational distribution data were nonquantitative and tended to have a number of false positives, while the wind and current data used in the computational procedure lacked detail or were impr...