Background Preterm infants are at great risk of bloodstream infection while receiving intensive care. The incidence of late-onset bloodstream infection varies between 10–60%. Diagnosis is currently made with a combination of clinical signs and laboratory investigations. The use of new molecular methods to improve diagnosis in a range of infections has been reported in other age groups. Objective To compare a molecular 16S rRNA assay to conventional laboratory methods (C-reactive protein (CRP) and blood culture) in the diagnosis of late-onset sepsis in preterm babies. Methods A single-centre, prospective observational study was conducted. All babies with suspicion of late-onset sepsis were included. Blood for the 16S rRNA assay was taken along with routine investigations for sepsis (blood culture, CRP). The decision to label a baby as being “septic” was based on clinical data and the assessment by a national nosocomial infection scoring system. Sepsis status was then compared to results of blood culture, CRP and the 16S rRNA assay. Diagnostic odds ratio, sensitivity, specificity, positive likelihood and negative likelihood ratios were calculated. Results Conclusion This assay, which enables generic detection of bacteria without reliance on culture, has superior sensitivity and specificity in the diagnosis of late-onset infection than conventional laboratory methods. These results indicate that molecular methods should have a place in the diagnosis of late-onset sepsis in preterm babies. Further study is required to determine the extent to which these methods could be utilised.