We prospectively examined whether febrile infants younger than 2 months of age who were defined as being at low risk for having bacterial infection could be observed as outpatients without the usual complete evaluation for sepsis and without antibiotic treatment. A total of 237 previously healthy febrile infants were seen at the Pediatric Emergency Room over 17 1/2 months. One hundred forty-eight infants (63%) fulfilled the criteria for being at low risk: no physical findings consisting of soft tissue or skeletal infections, no purulent otitis media, normal urinalysis, less than 25 white blood cells per high-power field on microsopic stool examination, peripheral leukocyte count 5000 to 15,000/mm3 with less than 1500 band cells/mm3. One infant appeared too ill to be included, and had sepsis and meningitis. None of the 148 infants at low risk had bacterial infections, versus 21 of 88 (24%) of those at high risk (P less than 0.0001); eight of 88 (9%) had bacteremia. Of the 148 infants classified as being at low risk for having bacterial infection, 62 (42%) were discharged to home, and 72 (49%) were initially observed for less than or equal to 24 hours and then discharged. Seventeen infants (11%) were hospitalized: in six, low risk became high risk; six had indications other than fever; and five because the study physicians could not be found. The 137 nontreated infants were closely observed as outpatients. The duration of fever was less than 48 hours in 42%, and less than 96 hours in 91%. All infants were observed for at least 10 days after the last examination. The fever resolved spontaneously in all infants but two, with otitis media, who were treated as outpatients. Our data suggest that management of fever in selected young infants as outpatients is feasible if meticulous follow-up is provided.