Accurate forecasting of dense fog in the Tallahassee, FL area is very challenging. This paper first will describe the climatology of fog in the area based on surface data and radiosonde observations. The occurrence of dense fog is related to parameters such as temperature, humidity, and winds. The Florida State University/Oregon State University Planetary Boundary Layer (PBL) model is run on selected dense fog and no fog nights to determine the extent to which the model provides useful forecast guidance. The paper then will describe how the PBL model output data are combined with the fog climatology to produce decision trees to improve dense fog forecasting in the area. A separate decision tree is prepared for each season. Individual meteorological parameters are combined and grouped ito three branches: a surface/radiosonde data branch, a PBL model output branch, and an advective fog/cloud/no PBL run branch. Thus, the forecaster has the option of using only observed data as guidance or of also utilizing the model output. The forecasting skill of the combined parameters in each branch is measured using the True Skill Statistic. Results show that the PBL model output branch generally produces slightly higher skill scores than the surface/radiosonde branch. The winter season decision tree exhibits the highest overall forecst skill, while the warm season decision tree has the lowest forecast skill. The paper will describe how local forecast offices can produce similar forecast guidance by running their own PBL model and establishing the statistical relationships that comprise the decision tree.