This study examined first-year attrition at a large, urban university in the United Kingdom, demonstrating the application of multilevel modeling to the issue of student attrition. A sample of 2,679 full-time, first-year students studying the 20 most common subject areas was identified. Students were divided into four groups depending on their entry route: entering after the 6th year of secondary school after obtaining sufficient Scottish school-leaving qualifications to enter the university in the 5th year; entering with English school-leaving qualifications (which require 6 years of secondary school); entering with Scottish school-leaving qualifications obtained in the year of their leaving secondary school (5th or 6th year); and entering with non-school-leaving qualifications, mainly gained at college. Students' subsequent first-year dropout rates were, respectively, 5.6 percent, 7.6 percent, 11.6 percent, and 18.2 percent. A multilevel random coefficient model was fitted to the data. Data from logistic regression analysis and multilevel analysis indicated that students studying certain subjects had significantly different withdrawal rates. There was little evidence of an interaction effect between subjects and entry routes. There were some significant differences among the withdrawal rates of students in the four entry route groups. (Contains 30 references.) (SM)