Of the 1,153 infants who completed monitoring by Aug 1, 1984, through our program at Massachusetts General Hospital, 76 infants had an initial apnea spell during sleep which was characterized by a change in tone and color, was unresponsive to repeated vigorous stimulation, and was terminated only after mouth to mouth resuscitation. The infants were hospitalized for observation and evaluation, and no cause could be identified. All were discharged on a home apnea or cardiorespiratory monitor, and subsequent episodes of apnea and/or bradycardia were reviewed. We grouped infants based on the intervention used to terminate subsequent episodes: Group 1, resuscitation; group 2, vigorous stimulation; and group 3, neither resuscitation or vigorous stimulation. There was no significant difference in clinical features or in the results of the initial evaluation in groups 1 and 2, compared with group 3. However, the mortality rate was significantly higher in group 1 (4/13) and group 2 (3/12) than in group 3 (3/51) (P less than .007). Siblings of victims of sudden infant death syndrome (n = 8) were at a significantly higher risk of an adverse outcome (two deaths and four resuscitations) than nonsiblings (P less than .02). A seizure disorder that developed during monitoring was associated with a high mortality (4/11 v 6/65, P less than .02). We conclude that these relatively rare infants who have sleep-onset apnea that responded only to resuscitation and have a subsequent similar episode or are siblings of victims of sudden infant death syndrome or develop a seizure disorder during monitoring have a very high risk of dying (31%, 25%, and 36% respectively).(ABSTRACT TRUNCATED AT 250 WORDS)