This paper tackles the relationship between text typology and machine translation. In this paper , the study is restricted to only one direction of translation that is from Arabic into English. It aims at clarifying the different results of machine translation renditions due to different text types. In theory, the paper gives a theoretical background on the topic. Then, practically, in order to test the hypothesis of the research which says that text typology affects the degree of appropriates of renditions by machine translation ,six functionally different texts (two of each informative, expressive and vocative texts) have been rendered from Arabic into English by machine translation (Google translate). Then the percentage of errors in these renditions has been calculated for each text by comparing machine translation with professional human one. The study concludes that that text typology affects the percentage of errors in MT in that fewer errors are noticed in vocative types compared with expressive and informative ones.