A procedure for using Guilford's structure of the intellect as the theoretical basis for a task analysis model is presented. It is reasoned that such a model would furnish a bridge between task analysis and test selection, and also a bridge between test selection and test validation. Such a mechanism might answer some of the Equal Employment Opportunity Commission (EEOC) criticisms of psychological testing because of the inherent content validity of the technique. In addition, the technique could be used to produce task analyses of both a job and training for that job. A comparison of the structural task analyses of the job and job training would expose discrepancies where abilities not required on the job are required for training. It might be then possible to modify training to bring the abilities required for training more in line with those required for the job, thus ultimately providing a bridge between task analysis through selection to training. (Author/RC)