This paper is directed to those persons interested in predicting concentrations downwind of platform-type structures associated with oil or gas facilities that operate on the Outer Continental Shelf. The specific purpose of this study was to determine the effect of platformtype structures on the dispersion of pollutant plumes and to assess the adequacy of the building wake algorithm included in the Offshore and Coastal Dispersion (OCD) Model. To meet the study objectives, a comprehensive wind-tunnel modeling study was conducted. Scale models of three typical oil platforms were positioned in an open-circuit wind tunnel and various source and meteorological conditions were simulated. Concentration and visual measurements were then obtained so that the dispersion characteristics could be quantitatively and qualitatively defined. Prior to conducting the platform wake evaluation, wind tunnel tests were conducted simulating two cases from tracer field experiments conducted in theGulf of Mexico. The simulations demonstrated that the wind tunnel can adequately simulate dispersion over water. Measurements of the horizontal and vertical dispersion coefficients in the wake of two different platform-type structures demonstrated that the platform can significantly affect dispersion. An equation that described the horizontal and vertical dispersion was of the form (i) (ii) where σ′(x) is the dispersion rate due to platform and ambient turbulence effects, σ (x) the dispersion rate due to ambient turbulence effects alone,A,B,C empirical constants,L a platform length scale,H b the platform height, andx 0 a virtual source distance. In the OCD model the horizontal and vertical dispersion coefficients are estimated using an equation of the following form: where σ0 is computed from an initial dilution radius and is a constant for a given building configuration. Predictions of σ′ using the OCD wake algorithm, a modified OCD algorithm and Equations i and ii were compared with results from the wind tunnel simulations. A comparison of observations and predictions showed that OCD generally underestimated the horizontal and vertical dispersion coefficients when platform wake effects are significant. Equationsi andii on the other hand gave predictions that agreed well with the wind tunnel observations. The agreement was significantly better for this algorithm (at the 95% confidence interval) than for the OCD and modified OCD wake algorithms.