Robust evaluation of public health interventions is required to ensure that interventions that lead to the greatest health benefit are adopted. However, traditional experimental evaluative designs are rarely possible for public health evaluation. Furthermore, alternative “quasi-experimental” designs are underused, are seldom covered in detail in epidemiology courses and are excluded from many guidelines and reviews. As a result population level health interventions have suffered from an “evaluative bias” whereby interventions not amenable to randomised control trials are often either poorly evaluated or not evaluated at all. Interrupted time series (ITS) analysis is one of the most powerful quasi-experimental designs for evaluating the effectiveness of population level health interventions. It is increasingly being used to evaluate the effectiveness of interventions ranging from clinical guidelines to national public health legislation. The basic design involves comparing the outcome of interest before and after an intervention, whilst accounting for any underlying trend. Nevertheless, ITS studies, like other quasiexperiments have more inherent threats to their internal validity than experimental designs, many of which have not been adequately addressed in the existing literature. Further guidance is needed on these threats and how they are best addressed in the design, application and appraisal of ITS studies. The overarching aims of this thesis are to improve the way that interrupted time series studies of public health interventions are designed in order to reduce the risk of bias and to make robust ITS designs more accessible to evaluators of public health interventions. This will be achieved through a range of methodological and applied studies using ITS designs.