Background: Rheumatic fever (RF) and its sequela rheumatic heart disease (RHD) occur at unusually high levels in New Zealand. A disease of social deprivation, RF now almost exclusively affects Māori and Pacific peoples. In 2011, the Government announced a goal to reduce the incidence of RF by two-thirds to 1.4 per 100,000 and make RF a rare occurrence for all by June 2017. The Ministry of Health identified improving RF surveillance as a major task to support this goal and consequently commissioned this project. This project aimed to describe the existing RF surveillance systems and identify potential surveillance improvements which would support more effective control and prevention of RF. It also aimed to estimate the likely true incidence and distribution of RF. A third aim was to assess the impact of certain factors on the extent and accuracy of case diagnosis, with a focus on echocardiography as a screening tool. Methods: A surveillance sector review framework was used to describe existing RF surveillance systems (the surveillance sector) and perform a gap analysis. Interviews with key informants were an important part of this process. Key informants were also asked about factors affecting the extent and accuracy of case diagnosis. A literature review was performed, describing RF surveillance systems and screening strategies throughout the world. Hospitalisation and notification data were obtained. Capture-recapture statistical analyses were performed to estimate the likely true incidence and distribution of RF. Data from four registers as well as Auckland register statistics were obtained and used to refine the estimates of RF incidence. The likelihood of matching individuals from one dataset to other datasets was calculated.Results: Intensive RF surveillance is rarely performed in developed countries outside of New Zealand and parts of Australia, mainly because RF is now very rare in these other countries. New Zealand needs a single, complete RF dataset from which accurate data can be obtained. A national register is likely to be extremely useful, both for strategy-focused surveillance (for example evaluating the impact of RF prevention programs) and control-focused surveillance (case management). As the notification database, EpiSurv, has most of the required features and is supported by the legal requirement to notify RF, it could be modified to act as a national register. It could also collect information about cases' exposures to known risk-factors. Increasing surveillance of intermediate outcomes and upstream hazards would also be beneficial. Over the period 1997-2011, it is likely there were between 2235-2337 new cases of RF, an average of 149 to 156 per year. Due to inconstant diagnostic procedures, the extent and accuracy of case diagnosis is likely to vary considerably. Conclusions: This thesis contains a set of recommendations to improve RF surveillance. Several of these would be relatively simple extensions of the existing systems and are likely to greatly increase the quality and usefulness of the surveillance data. Steps to address other important gaps could be approached as short-to-medium term goals. While the current RF surveillance sector is flawed, its gaps are not insurmountable.