Summary Background Our aim was to identify which clinical features have value in confirming or excluding the possibility of serious infection in children presenting to ambulatory care settings in developed countries. Methods In this systematic review, we searched electronic databases (Medline, Embase, DARE, CINAHL), reference lists of relevant studies, and contacted experts to identify articles assessing clinical features of serious infection in children. 1939 potentially relevant studies were identified. Studies were selected on the basis of six criteria: design (studies of diagnostic accuracy or prediction rules), participants (otherwise healthy children aged 1 month to 18 years), setting (ambulatory care), outcome (serious infection), features assessed (assessable in ambulatory care setting), and sufficient data reported. Quality assessment was based on the Quality Assessment of Diagnostic Accuracy Studies criteria. We calculated likelihood ratios for the presence (positive likelihood ratio) or absence (negative likelihood ratio) of each clinical feature and pre-test and post-test probabilities of the outcome. Clinical features with a positive likelihood ratio of more than 5·0 were deemed red flags (ie, warning signs for serious infection); features with a negative likelihood ratio of less than 0·2 were deemed rule-out signs. Findings 30 studies were included in the analysis. Cyanosis (positive likelihood ratio range 2·66–52·20), rapid breathing (1·26–9·78), poor peripheral perfusion (2·39–38·80), and petechial rash (6·18–83·70) were identified as red flags in several studies. Parental concern (positive likelihood ratio 14·40, 95% CI 9·30–22·10) and clinician instinct (positive likelihood ratio 23·50, 95 % CI 16·80–32·70) were identified as strong red flags in one primary care study. Temperature of 40°C or more has value as a red flag in settings with a low prevalence of serious infection. No single clinical feature has rule-out value but some combinations can be used to exclude the possibility of serious infection—for example, pneumonia is very unlikely (negative likelihood ratio 0·07, 95% CI 0·01–0·46) if the child is not short of breath and there is no parental concern. The Yale Observation Scale had little value in confirming (positive likelihood ratio range 1·10–6·70) or excluding (negative likelihood ratio range 0·16–0·97) the possibility of serious infection. Interpretation The red flags for serious infection that we identified should be used routinely, but serious illness will still be missed without effective use of precautionary measures. We now need to identify the level of risk at which clinical action should be taken. Funding Health Technology Assessment and National Institute for Health Research National School for Primary Care Research.