Recognizing textual entailment is a challenging problem and a fundamental component of many applications in natural language processing. We present a novel framework for recognizing textual entailment that focuses on the use of syntactic heuristics to recognize false entailment. We give a thorough analysis of our system, which demonstrates state-of-the-art performance on a widely-used test set.