As technology is advancing, larger and denser devices are being manufactured with shorter time to market requirements. Identifying and resolving problems in integrated circuits (ICs) are the main focus of the pre-silicon and post-silicon debug process. As indicated in the International Technology Roadmap for Semiconductors (ITRS), post-silicon debug is a major time consuming challenge that has significant impact on the development cycle of a new chip. Since it is difficult to acquire the internal signal values, conventional debug techniques typically involve performing a binary search for failing vectors and performing mechanical measurement with a probing needle. Silicon debug is a labor intensive task and requires much experience in validating the first silicon. Finding information about when (temporal) and where (spatial) failures occur is the key issue in post-silicon debug. Test vectors and test applications are run on first silicon to verify the functionality when it arrives. Scan chains and on-chip memories have been used to provide the valuable internal signal observation information for the silicon debug process. In this dissertation, a scan-based technique is presented to detect the circuit misbehavior without halting the system. A debugging technique that uses a trace buffer is introduced to efficiently store a series of data obtained by a two dimensional compaction technique. Debugging capability can be maximized by observing the right set of signals to observe. A method for an automated selection of signals to observe is proposed for efficient selection. Investigation in signal observability is further extended to signal controllability in test point insertion. Noble test point insertion techniques are presented to reduce the area overhead for test point insertion. / text