This book is based on the author's PhD thesis which was selected during the 1993 ACM Doctoral Dissertation Competition as one of the three best submissions. This monograph investigates the problem of selecting adequate models for reasoning about physical systems and applications to engineering problem solving. An elegant treatment of both the theoretical and practical sides are presented: the problem is precisely formalized, its computational complexity is analyzed in detail, and an efficient algorithm for finding adequate models is derived; on the practical side, a methodology for building systems that automatically construct adequate models is provided, and implementational aspects and tests are described.