Software functional test is a challenge faced by developers for a long time. The growing complexity of computing systems turns this challenge even greater. Model-based testing is a trend pointed out by the academia and the industry as a possible solution to this matter. Inspired by this paradigm, this dissertation depicts a research made on the use of grammars as functional test models. Grammar models can capture concepts and behaviors of a system and its environment at a level of abstraction according to the test goal. They also can be applied to describe functional test cases and guide the execution of the generated test cases against a system under test. The result of this execution, represented as a verdict, reveals the system conformity with its requirements and specifications. In order to explore grammar models potential, this work defined a systematic way to generate and execute a mass of tests. This solution allowed the implementation of different test strategies. It also assisted test adjustment to requirements change and promoted existing tests reuse. As a side-effect of this study, a functional test process was developed and the supporting architecture introduced here may be reused or extended by future functional test solutions.