This thesis presents three chapters on the subject of unemployment persistence. Two of the chapters are empirically focussed and the other is a purely theoretic work. Unemployment persistence is defined as the existence of serial correlation in individual employment outcomes. The first chapter finds evidence for unemployment persistence among men and women in the Australian youth labour market. Individual labour market dynamics are analysed using the Australian Longitudinal Survey. The analytic framework used is a Random Effects Probit model, incorporating lagged employment status as an explanatory variable status. Results support a “scarring” effect of unemployment upon individuals’ future employment prospects. The second chapter provides decision-theoretic foundations for unemployment persistence, based upon heterogeneous intrinsic productivity among workers. A representative firm is assumed to receive an imperfectly precise signal of worker ability every period, and re-forms its beliefs every period using a Bayesian updating method. A model of the dynamic behaviour of optimal employment decisions by the firm is constructed. It is shown that under certain circumstances workers of all productivities may be “scarred” in the eyes of the firm by past unemployment, due to the firm’s being unwilling to hire from an unemployment pool of dubious quality. The third chapter presents a detailed investigation into how to measure unemployment persistence within the UK. The chapter presents several modelling strategies capable of being used to analyse panel data of a binary nature, and discusses how to decide which methods are most appropriate in particular environments. Panel data on men from the British Household Panel Survey are used to estimate a structural state dependence equation in employment status, where lagged employment status is used as an explanatory variable. Particular attention is given to controlling for unobserved heterogeneity between individuals. The empirical results indicate strong evidence of unemployment persistence.