Abstract
During the academic year 1995–1996 a series of Mathematics Support tutorials were held at King's College London for first-year undergraduates in engineering and science. The students were chosen as a result of their performance on a mathematics pre-test held during their first week at university. The test had been designed and used more than a decade ago for use with pre-university students. A comparison between the 1980s and 1990s performance on the test is carried out and reasons for changes suggested. A further comparison is made between the engineering under graduates' performance on the pre-test and their scores on their later end-ofsemester mathematics examination. This reveals a pattern indicating that an all round understanding of graphs on the pre-test predicted success on the later mathematics examination, despite an almost total lack of graph questions on that examination. Reasons for this pattern are suggested and discussed.