As circuits scale to ever smaller feature sizes, the types of defects that occur become increasingly complex and difficult to model. They generally do not behave like the faults considered during test pattern generation and therefore must be fortuitously detected by the pattern set. This mismatch of faults and defects also poses significant problems for diagnosis. This presentation describes a preliminary investigation into the use of mandatory conditions for the detection of both faults and defects with applications to defect diagnosis. It will show that multiple site observations and good excitation balance are essential not only for adequate fortuitous defect detection - they are also necessary for the determination of the mandatory conditions and implications that accompany those detections, and thus enable the diagnosis of these fortuitously detected and unmodeled defects as well.