We present a feasibility study to protect smart card software against fault-injection attacks by means of binary code rewriting. We implemented a range of protection techniques in a link-time rewriter and evaluate and discuss the obtained coverage, the associated overhead and engineering effort, as well as its practical usability.