One basic intent of feature-based modeling is to provide objects and operators which support the implementation of a complete and consistent description of a part in terms of its shape and related functional and technological information. Although feature-based methodology has matured to the point that results are being incorporated into commercial CAD systems, several problems still remain open. A key problem in feature-based modeling is how to maintain the consistent correspondence between the geometrical description of a feature and its related functional meaning (semantics) during the entire modeling process.Uncontrolled feature shape modification due to geometrical operations among aggregated features can affect the correct functionality of a modeled part. Within described work we propose to approach the problem of controlling feature semantics by using what we call self validation features: an entity concept developed to implement feature specifications replenished with self validation capabilities, Along with this new approach the traditional feature definition is extended in order to include a set of rules that allows for feature instances to control the consistency of their shape in respect to functionality associated. A feature-based modeling operation is then executed by computing within three steps as follows: feature instantiation, feature location and validated feature aggregation. A prototype testbed based on a self validation entity concept has been implemented integrating an in house developed feature system with a commercially available geometric modeling kernel, In order to realize a tight integration between feature semantics and shape representation an interfacing mechanism, based on an entity monitor, has been studied. Its lunctionality is partly supported by the attribute handling system of the integrated geometric modeling kernel. In the example section of the reported experimental work a selection of validation rules peculiar to some semantics aspects are evaluated and discussed.