Empirical physics-based methods for predicting protein and crystal structures from potential functions, without ancillary aids such as secondary structure prediction, homology modeling, threading, fragment coupling, or use of space-group information, have evolved to a point where reasonable agreement between predicted and experimental structures can be expected. Recently-introduced blind tests by CASP (Critical Assessment of Methods of Protein Structure Prediction), for judging protein structure predictions, and by the Cambridge Crystallographic Data Centre, for judging crystal structure predictions, have provided objective evaluations of the relevant computational procedures. We have participated in both types of tests which have indicated the strengths and weaknesses of our procedures, thereby providing information about the directions in which to increase the reliability of our techniques. Our procedures, and results thereof, will be described.