This thesis describes study designs for the robust evaluation of complex patient safety interventions. Fundamentally, study designs available to measure the effectiveness of patient safety interventions fall into two categories – those that use contemporaneous controls, and those that do not. A review of the recent literature (245 citations) revealed that most studies were single-centre (63%), and the majority of these did not use contemporaneous controls (84%); whilst in multi-centre studies (37%) the number of studies using contemporaneous controls (49%) equalled the number of studies that that did not (51%). Studies that do not use contemporaneous controls dominate the literature, but they are weak and subject to bias. The thesis further discussed a case-study, as an exemplar for the evaluation of a highly complex patient safety intervention – the Safer Patients Initiative (SPI), which sought to generically strengthen hospitals, whilst improving frontline activities. The evaluation was a before and after study, with contemporaneous controls. It used mixed-methods, so that the triangulation of a one type of research finding could be reinforced when corroborated by the finding of another type. Uniquely, it also, compared the rates of change across control and SPI hospitals – an approach referred to as the “difference-in-difference” method.