We propose Bilingually-constrained Recursive Auto-encoders (BRAE) to learn phrase embeddings (compact vector representations for phrases), which can distinguish the phrases in different semantic meanings. The BRAE is trained with the objective to minimize the semantic distance of translation equivalents and maximize the semantic distance of nontranslation pairs. The learned model can embed any phrase semantically in two languages and can transform semantic space in one language to the other. We evaluate the BRAE on two end-to-end SMT tasks (phrase table pruning and translation hypotheses reranking) which need to measure semantic similarity between a source phrase and its translation candidates. Extensive experiments show that the BRAE is spectacularly successful in these two tasks.