This paper presents examples of various ways of utilising for research purposes large nationwide registers, originally collected for administrative purposes. Merging registers with information on sociodemographic factors can help understand trends in work disability outcomes, such as sickness absence and disability retirement. Quasi-experiments can be designed utilising large register data to look at the effectiveness and cost-effectiveness of interventions on a societal level, such as legislative changes. Examples of such interventions include the introduction and amendment of laws determining the use of part-time sick leave in Finland and the so-called 30-60-90 rule, defining check-points for action for the occupational service provider, the employer and the employee in situations of prolonged sickness absence. Advanced analytical techniques, for example the construction of the propensity score, to increase comparability of studied groups, are especially feasible in large register data. We recommend using nationwide administrative register data for research purposes, where such data are available.