To include a specific antibiotic in the empiric therapy, it is necessary to predict when a nosocomial pneumonia (NP) is caused by methicillin-resistant Staphylococcus aureus (MRSA). We have developed a model for the prediction of the probability of a NP being caused by MRSA, when the carrier status and the microbiological diagnosis are unknown. A retrospective case-control study (1999-2005) was designed. A univariate and multivariate logistic regression was performed to identify the risk factors for suffering a NP due to MRSA. Demographic factors, related to hospitalization, immunosuppression or neutropenia, to medication and severity were included. Three hundred and sixty three patients (121 cases and 242 controls) were studied. The final model of multivariate logistic regression included an age>14 years (OR 7.4, CI 95% 1.5-37.4, P<.015), NP appearance>6 days after admittance (OR 4.1, CI 95% 2.4-7,1, P<.001), NP development excluding summers (OR 2.5, CI 95% 1.2-5.2, P<.015), respiratory diseases (OR 4.9, CI 95% 1.5-15.8, P<.007) and multilobar involvement (OR 4, CI 95% 2.3-7.2, P<.001).The probability of developing a pneumonia due to MRSA was studied for each of the possible combinations and subsequently classified in minor and major criteria. MRSA coverage should be included in the empirical treatment of NP when: a) an adult patient (>14 years old) presents, at least, 2 major criteria or 1 major criterion together with 2 minor criteria, and b) a patient <14 years-old has 2 major criteria as well as 2 minor criteria.