OBJECTIVES: Resistance to antibiotics is a serious public health problem with an increasing economical impact. A wide knowledge of the evolution and characteristics of these bacterial is thus critical to manage and develop treatment and prevention strategies. The main goal of this study is assessing probabilities of MRSA infection in the Comunitat Valenciana according to hospital and nonhospital setting, gender, age and specimen. METHODS: Data are supplied by RedMIVA, the microbiological surveillance network that collects information from the public microbiology laboratories. A total of 2034 determinations are recorded from January to June 2006. The statistical model is a Generalised Additive Model with gender, hospital and nonhospital setting and specimen as factors, and the age as a covariate included in the model as a nonlinear term with splines RESULTS: The expected estimated probability of MRSA is less in women than in men in all the ages. The exception occurs in hospitalised women aged 20 to 40 which exhibit significant fluctuations. Urine and oftalmological specimen produce always the largest and smallest probabilities. The fitted model allows the estimation of probabilities of MRSA for any combination of the factor levels and age. For example, in the case of urine specimen, the posterior expected probability in hospitalised (nonhospitalised) men with ages 20, 40, 60 and 80 is 0.3974 (0.2236), 0.5127 (0.2608), 0.6753 (0.5062) and 0.8094 (0.73332), respectively. In the group of urine specimen supplied by hospitalised (nonhospitalised) women the results are 0.3354 (0.4720), 0.5400 (0.009), 0.6716 (0.4473) and 0.6463 (0.5458) for ages 20, 40, 60 and 80, respectively. CONCLUSION: We have fitted an stochastic model for learning about MRSA that essentially captures its behaviour in the population. The probabilities of MRSA are, in general, higher for men than women. They increase as people get old but its difference decreases with age.