The nature of vocabulary selection has long been problematical for ESP practitioners such as those working in the field of Medical English, a genre of interest to the readers of this journal. In an attempt to quantify the vocabulary of medical research articles, this paper reports on a computer corpus comprising articles covering a range of specialized medical fields. The occurrences of the most frequent items on this medical list are compared with their corresponding totals on the Cobuild corpus of general English. Considerable differences are found between the two frequency lists, and the most important of these are discussed. It is suggested that a general English frequency list is inadequate as a source for vocabulary selection for a Medical English curriculum, and that a subject-specific medical word count has far greater applicability.