We reviewed etiology and outcome of consecutive neonates admitted to a neonatal unit for investigation of parent-reported fever (116 neonates over 24 months). Tympanic temperature was measured at the emergency department (Te) and core temperature at the neonatal unit (Tn). Microbials were isolated in 27 patients (23%); Te and Tn were both <38°C in 13 (48%) of the 27 patients. Microbial isolation was associated with older median age (16.7 vs. 8.0 days, p = 0.004), empirical antibiotic commencement (p = 0.0003) and longer hospital stay (median 8 vs. 4.0 days, p = 0.004). Compared with respiratory viral infection, patients with bacteremia had high C-reactive protein (p = 0.005) and likely to have comorbidity of meningitis (p = 0.077). Te ≥38°C had the highest sensitivity, positive likelihood ratio and positive and negative predictive ratios for bacteremia. Parent-reported fever was associated with a 3% incidence of meningitis, 6% of bacteremia and 9% of urinary tract infection. The majority of neonates with parent-reported fever do not have serious bacterial infection. Nevertheless, recommendations about threshold of antibiotic initiation are difficult, and empirical systemic antibiotic coverage must be commenced in those neonates with Te ≥38°C or elevated C-reactive protein.