ABSTRACT This report describes an investigation of the practical applicability and accuracy of the OCD model for simulating air pollution concentrations from offshore emission sources such as oil drilling platforms. The OCD computer code (Version 2) was furnished to SRI International by the American Petroleum Institute in March 1985. The behavior of the model has been examined as the input variables are assigned values that encompass their normal ranges@ and the sensitivity of the OCD model results to the variables has been determined. The calculated concentrations at or near the shoreline are most sensitive to overwater turbulence intensity@ wind speed@ and mixing height. If turbulence intensity data are not available@ the air-minus-sea temperature difference takes their place. Farther inland@ overland stability becomes increasingly important. Using air-minus-sea surface temperature difference to characterize the stability of the atmosphere@ the accuracy of the model has been evaluated. This was done by comparing computer simulations to onshore tracer measurements made during the Pismo Beach@ California@ and Cameron@ Louisiana@ dispersion experiments. The accuracy in simulating onshore concentrations was found to be rather low when predicted and observed values were paired in time and space. The associated correlation coefficients were between 0.39 and 0.69. Also@ there was a distinct tendency for underprediction. However@ this evaluation procedure@ using pairs of concurrent predicted and observed concentrations@ is a stringent test of model performance. Evaluations based on totally unpaired comparisons (often used in regulatory applications) may provide different results. Finally@ results are presented of an assessment of the availability and costs of providing special meteorological measurements for input to the OCD model.