Unemployment and the duration of stay in this labour market state have motivated many empirical studies in the last three decades. The main focus of the empirics, up until recently, has been to model the total duration of unemployment, defined by the termination of a job and the commencement of a new labour market state. This new labour market state is most often defined by the start of a new job. However, many labour markets function within a system of wage replacement benefits for unemployed workers via state-run Unemployment Insurance (UI) programmes. More recent studies have focused on the period of insured unemployment in relation to specific programme rules such as wage replacement level (the benefit level) and maximum weeks of benefit entitlement. This thesis takes advantage of rule changes to the benefit level and maximum entitlement within the Canadian UI programme in the early 1990s to examine their impact on UI durations. A random sample of UI claims filed by male claimants from thirteen geographic regions within the four provinces of Atlantic Canada was selected. Unemployment Insurance durations are defined both by the last week of UI benefit receipt and the first week of active labour market activity while on claim. The analysis uses non-parametric, semi-parametric and parametric estimation techniques in obtaining covariate and baseline hazard influences on the duration of UI claims. The estimation results indicate that there are statistically significant benefit and entitlement effects as well as UI duration differences by region, occupation and industry. Unique UI durations were also revealed for exit events defined by the start of part-time earnings, full-time earnings and the start date of training while on a UI claim. Overall, programme changes which have reduced programme generosity have shortened UI durations primarily by reducing the maximum number of weeks ofUI entitlement