A gradient-analysis-based model and grid-based map are presented that use the potential vegetation zone as the object of the model. Several new variables are presented that describe the environmental gradients of the landscape at different scales. Boundary algorithms are conceptualized, and then defined, that describe the environmental boundaries between vegetation zones on the Olympic Peninsula, Washington, USA. The model accurately predicted the vegetation zone for 76.4 percent of the 1,497 ecoplots used to build the model. Independent plot sets used to validate the model had an accuracy of 82.1 percent on national forest land, and 71.8 percent on non-national-forest land. This study demonstrated that a model based on boundary algorithms can be an alternative to regression-based models for predicting landscape vegetation patterns. Potential applications of the model and use of the environmental gradients to address ecological questions and resource management issues are presented.