The paper describes some tractability conditions for classes of CNFs and their influence on the computational and algorithmic complexity of CNF minimization within these classes