BACKGROUND: The study of severe maternal morbidity survivors (near miss) may be an alternative or a complement to the study of maternal death events as a health care indicator. However, there is still controversy regarding the criteria for identification of near-miss maternal morbidity. This study aimed to characterize the near miss maternal morbidity according to different sets of criteria. METHODS: A descriptive study in a tertiary center including 2,929 women who delivered there between July 2003 and June 2004. Possible cases of near miss were daily screened by checking different sets of criteria proposed elsewhere. The main outcome measures were: rate of near miss and its primary determinant factors, criteria for its identification, total hospital stay, ICU stay, and number and kind of special procedures performed. RESULTS: There were two maternal deaths and 124 cases of near miss were identified, with 102 of them admitted to the ICU (80.9%). Among the 126 special procedures performed, the most frequent were central venous access, echocardiography and invasive mechanical ventilation. The mean hospital stay was 10.3 (+/- 13.24) days. Hospital stay and the number of special procedures performed were significantly higher when the organ dysfunction based criteria were applied. CONCLUSION: The adoption of a two level screening strategy may lead to the development of a consistent severe maternal morbidity surveillance system but further research is needed before worldwide near miss criteria can be assumed.