This paper examines the persistence of under-employment amongst UK higher education graduates. For the cohort of individuals who graduated in 2002/3, micro-data collected by the Higher Education Statistical Agency, are used to calculate the rates of "non-graduate job" employment 6 months and 42 months after graduation. A logit regression analysis suggests the underemployment is not a short-term phenomenon and is systematically related to a set of observable characteristics. It is also found that under-employment 6 months after graduation is positively related to under-employment 42 months after graduation, which is consistent with the view that the nature of the first job after graduation is important in terms of occupational attainment later in the life-cycle.