OBJECTIVES: Estimate the revision rates and life time of knee and hip prosthesis in the state of São Paulo, Brazil.   METHODS: A longitudinal analysis (2008 to 2014) of an anonymized government administrative database (Inpatient Information System - SIH/DATASUS) was performed. The study cohort comprised knee and hip arthroplasties. Patients were uniquely identified by the following parameters: date of birth, postal code and gender. Deterministic registry matching was applied. The analysis considered patients with a primary arthroplasty between 2008 and 2011. The follow up to identify revision procedures were until 2014. Exclusion criteria were: (i) more than five procedures during follow-up period; (ii) individuals with the first or only procedure registered as revision; (iii) revision with less than six months were considered as bilateral. Procedures were classified as cemented or non-cemented. Outcomes assessed were revision rates and prosthesis’s lifetime. Descriptive statistics [i.e., average and proportions] summarized studied data.  RESULTS: