In order to study the occurrence and positive predictive value of history and physical examination findings suggestive of serious illness in ill-appearing and well-appearing febrile children, 103 consecutive children aged less than or equal to 24 months with fever greater than or equal to 38.3 degrees C were evaluated from July 1, 1982 to Nov 24, 1982. Patients were initially classified by an attending physician (A) as to whether they appeared ill (Yale Observation Scale score greater than 10) or well (scale score less than or equal to 10). The history was then taken by two attending physicians (A and B) and a resident; the physical examination was performed by attending physician B and the same resident. As history and physical examination findings were elicited, they were scored as to whether they did or did not suggest a serious illness. Serious illness was defined as the presence of a positive laboratory test. Ill-appearing patients had a significantly greater (P less than .001, Fisher's exact test) occurrence of physical examination findings suggesting serious illness (14 of 22, 64%) than well-appearing children (12 of 81, 15%). The positive predictive values of abnormal physical examination findings for serious illness in ill-appearing (11 of 14, 79%) and well-appearing children (3 of 12, 25%) were significantly different (P = .02 by Fisher's exact test). The trends for abnormal history findings in ill-appearing and well-appearing children were similar to those for abnormal physical examination findings but did not achieve statistical significance. The results, indicating an important interaction between a febrile child's appearance and physical examination findings, are discussed in terms of probability reasoning in clinical decision making.