To determine whether significant historical differences distinguish the near-miss for Sudden Infant Death from the infants who died of SIDS, we analysed the histories and clinical data from two groups of infants seen in our University Hospital and from collaborative research group. The data were obtained with the use of a standardised questionnaire and consultation of all available medical data. Sixty-five infants were identified as near-miss for SIDS after they had suffered a severe cardiorespiratory incident during sleep for which no cause could be found despite a complete medical examination. After an autopsy had failed to reveal a cause for the unexpected death 95 cases of SIDS were retained in the study. A series of 353 variables were collected from the parents, the gynaecologists, neonatologists and attending physicians. After statistical analysis, only 15 of the 353 items studied significantly differentiated between the two groups. A step-wise discriminant analysis performed on these items led to the identification of six independent variables: the time of the incident; the circumstances leading to the observation of the child; the child's sleep position; previous minor intestinal problems; the size of the family and the mother's coffee consumption. Most variables indicate that the near-miss infants were discovered and rescued earlier than the infants who died. No other historical information appeared significantly to differentiate between the two groups of infants. These data need confirmation from a prospective epidemiological survey.