Salmonellosis is an important cause of foodborne human gastroenteritis in most European countries, and pork contributes significantly to the disease burden caused by Salmonella infections. A traditional, and very popular, pork product in Denmark is fried meatballs. Danish meatballs are typically made with ground pork as the main ingredient. They are consumed as a component in hot meals but are also widely used as filling in cold sandwiches. Because of their popularity, pork meatballs are often part of the product range in Danish catering settings. As illustrated by the fact that 55 of 77 reported outbreaks in Denmark in 2010 were associated with outside-the-home settings, such as restaurants, canteens, hotels, schools, shops, institutions and sport events (Anonymous 2011), food prepared outside the home is a significant source of foodborne illness. In the present study, Quantitative Microbiological Risk Assessment (QMRA), following the Codex Alimentarius Principles and using the modular process risk model (MPRM) methodology, was used as the tool to investigate the fate of Salmonella during processing of pork meatballs from the reception of whole pork cuttings, through processing, until the point of consumption in a catering unit, e.g. worksite or school canteens. Appropriate transfer, growth and inactivation models developed specifically for Salmonella spp. in fresh meat, preferably pork, were applied to evaluate different scenarios for a food processing line constructed from two observational studies. Applying the proposed QMRA model (MANUSCRIPT II), risk estimates of illness from meatballs processed by the catering sector per year in Denmark were obtained (no cases applying a core temperature >71°C, 27 cases applying a core temperature of 65°C, 69 cases without applying a final heat treatment in oven). When comparing with recent epidemiological data from 2011 (44 –131 reported cases) for Salmonella cases in Denmark related to pork (Anonymous, 2012), these estimates are considered reasonable and realistic. Core temperatures higher than 71°C were found sufficient to inactivate Salmonella in meatballs, indicating that the recommendation from the Danish Food Authorities of 75°C is sufficient to eliminate this pathogen. Survival as a result of ineffective heating and growth of Salmonella during cooling had relevant impact on the risk estimates and, therefore, these processing steps should continue to be considered critical for meatballs safety. However, more knowledge about the critical limits was obtained, e.g. cooling of meatballs at temperatures lower than 24°C for periods shorter than three hours does not allow growth of the remaining Salmonella cells. In addition, prevalence and concentration of Salmonellahad higher impact on the risk estimates, thereby justifying the efforts to keep these factors low. The flexible structure of the QMRA model allows scenario analysis, which consequently makes it possible to expand the use of the QMRA model from pathogen outbreak investigation to product development and other food safety evaluation activities, e.g. efficiency of guidelines and establishment of critical limits. Two of the predictive models used to build the QMRA model (MANUSCRIPT II) were also developed in this thesis: - The transfer of Salmonella during grinding of pork was successfully modelled in processing of up to 110 pork slices corresponding to 21 kg meat (PAPER I). This model includes the pieces of meat that are contaminated before grinding and it gives clear explanations of all the parameters involved that combined give an overview of the dynamics of a grinding process. The structure of the model, and particularly its ability to predict the tailing phenomenon of low contaminated portions, seems relevant for different cross contamination processes. - Effect of pork natural microbiota on growth of Salmonella was modelled and predicted during storage of ground pork at temperatures between 4°C and 38°C (MANUSCRIPT I). Continued growth of Salmonella after the natural microbiota had reached their max. population density was observed. This effect was described well by the complex LotkaVolterra species interaction model as well as a new expanded Jameson-effect model but not by the classical Jameson-effect model. As the new expanded Jameson-effect model is more simple and practical to use, it may be preferred over the Lotka-Volterra model. Performance of both models was, however, temperature dependent, presenting good results at temperatures from 15.1 to 20.2°C. The combination of data from observational studies, models specifically developed studying transfer and growth of Salmonella in pork (PAPER I and MANUSCRIPT I), and literature data related to Salmonella in different meat matrices resulted in a new approach that may improve the quality of estimates in risk assessments related to Salmonella in pork processed at the catering sector.