For decades, many investigators have attempted to identify clinical or laboratory markers that can accurately differentiate severe bacterial from self-limiting viral infections in young children with fever without source. Unfortunately, no perfect marker has been discovered so far. Many guidelines recommend white blood cell count as a screening marker in fever without source, whereas compelling evidence in the literature emphasizes the superior characteristics of C-reactive protein and procalcitonin. One way to improve predictive value is the combination of prediction rules of different tests for clinical and laboratory markers. Several clinical decision rules, reviewed in this article, have been suggested but seem to be difficult to implement in practice due to their complexity. Recently, procalcitonin, C-reactive protein and urinary dipstick were combined in a simple risk index score that displayed promising predictive value in severe bacterial infections in children. Ultimately, impact analyses still have to be performed to show improved quality of care in this setting.