The compositions or essay tests of second language learners are difficult to mark and evaluate. It has been suggested that the only way to achieve objectivity is to abolish all criteria except formal correctness, as it is the only one that can be measured and statistically evaluated. Attempts have been made to establish other criteria. A varied vocabulary ought to be as important as formal correctness, which on its own is quite inadequate. In this study, in an attempt to measure success, the lexical content of essays written in English by Swedish students has been examined. An attempt is made to apply certain theoretical methods of evaluation to the essays and to use the results for two main purposes: (1) to compare the lexical texture of Swedish students' work with that of native speakers' writing on the same subject, and (2) to see to what extent spontaneous evaluation by a native speaker can be supported by the theoretical results. The quality of the essay was then put into relation to other factors, the length of the essay and the number and predominant types of error. (Author/CLK)