Description: This article discusses metadata records translation. Abstract: In this paper, performance of online translation systems including Google, Systran and Bing on translating metadata records derived from the digital library- Portal to Texas History- is manually evaluated using four measures: Fluency, Adequacy, Incorrect Translation, and Missing Translation. The authors propose exploring multi-engine machine translation for improving the quality of translation and point out three possible strategies of implementing multilingual information access in digital libraries applying machine translation of metadata records.