This article examines the effects of transitory skill mismatch in a matching model with heterogeneous jobs and workers. In our model, some highly-educated workers may accept unskilled jobs for which they are over-qualified but are allowed to engage in on-the-job search in pursuit of a better job. We show that this feature has relevant implications for the set of potential equilibria, the unemployment rates of the different types of workers, the degree of wage inequality, and the response of the labour market to shifts in the demand and supply of skills.An examination to select 175 postal workers for the region of Madrid yesterday gathered 15,570 candidates. Among them, 53% were college graduates (ÔlicenciadosÕ or ÔdiplomadosÕ) while the required educational attainment was upper secondary education (Ôgraduado escolarÕ) or an equivalent level of vocational training.(EL PAI´S, 23rd March, 2002).As the excerpt from a Spanish newspaper illustrates, mismatch between the skill requirements of jobs and the educational attainments of workers can be a pervasive feature in some labour markets. In this article, we study this phenomenon in a matching model with heterogeneous jobs (skilled and unskilled) on one side and heterogeneous workers (highly and less educated) on the other side of the market. Since the matching technology is imperfect, the highly educated workers may end up in unskilled jobs for which they are over qualified. A key element in our analysis is that mismatched workers are allowed to keep the option of moving to better jobs through on the job search (henceforth, OTJ search). Skill mismatch has therefore a transitory nature in our economy leading to job to job (hereafter, JTJ) transitions which are shown to have relevant implications for the composition of jobs, unemployment, wages and the reaction of the labour market to shifts in the demand and supply of skills.Labour economists have long recognised the importance of JTJ flows but it is only recently that the literature on equilibrium unemployment has started to explore its implications systematically.1 Our article contributes to this stream of research by * We are grateful to Steve Pischke (the Editor) and two anonymous referees for thoughtful comments and suggestions. We have also benefited from comments on ealier drafts of this article by James Albrecht, Pieter Gautier, Barbara Petrongolo, Robert Shimer, Manuel Santos and participants at various seminars. We would also like to thank Maite Blázquez and Juan R. García for excellent research assistance. The usual disclaimer applies. Financial support from the Consolider-Ingenio 2010 (Ministerio de Educaci on) and Excelecon (Comunidad de Madrid) projects is gratefully acknowledged.1 Broadly speaking, the literature on OTJ search can be divided in two strands. One strand uses models in the vein of Burdett and Mortensen (1998) to explain how OTJ search may give rise to wage differentials among identical workers; see, e.g., Mortensen (2003) for an excellent overview of this literature. The second strand inco...