We present a phrase-based extension to memory-based machine translation. This form of example-based machine translation employs lazy-learning classifiers to translate fragments of the source sen-tence to fragments of the target sentence. Source-side fragments consist of variable-length phrasesin a local context of neighboring words, translated by the classifier to a target-language phrase. Wecompare three methods of phrase extraction, and present a new decoder that reassembles the trans-lated fragments into one final translation. Results show that one of the proposed phrase-extractionmethods—the one used in Moses—leads to a translation system that outperforms context-sensitiveword-based approaches. The differences, however, are small, arguably because the word-based ap-proaches already capture phrasal context implicitly due to their source-side and target-side contextsensitivity.