BACKGROUND: Coagulase-negative staphylococci (CoNS) are common contaminants in blood cultures (BC). A prospective study of patients with ≥ 2 blood culture sets and at least 1 positive CoNS BC was performed to develop an algorithm to assist in determining the clinical significance of CoNS bacteraemia. METHODS: A single reviewer examined the medical records of patients with CoNS bacteraemia (January-June 2010). The determination of clinical significance was made according to CDC/NHSN (US Centers for Disease Control and Prevention/National Healthcare Safety Network) criteria. To explore risk factors associated with clinical significance, a multivariate analysis was performed. The performances of various algorithms were then compared. An algorithm to assist in determining clinical significance was developed. RESULTS: Two hundred and sixty-nine cases were included; 97 (36%) were considered clinically significant bacteraemia (CSB). Predictors of CSB in the multivariate analysis were: time to positivity < 16 h (odds ratio (OR) 4.540, 95% confidence interval (CI) 1.734-11.884), identification of Staphylococcus epidermidis (OR 4.273, 95% CI 2.124-5.593), central venous catheter (OR 4.932, 95% CI 2.467-9.858), > 2 CoNS-positive bottles from different BC sets (OR 1.957, 95% CI 1.401-2.733), and Charlson score ≥ 3 (OR 2.102, 95% CI 1.078-4.099). The algorithm with best sensitivity (62%) and specificity (93%) for determining clinical significance of CoNS included Charlson score ≥ 3, Pitt score ≥ 1, neutropenic patients, presence of central venous catheter, identification of S. epidermidis, and time to positivity < 16 h. The positive predictive value was 83% and the negative predictive value was 81% (likelihood ratio 8.87). CONCLUSION: The use of this algorithm could potentially reduce the misclassification of nosocomial bloodstream infections and inappropriate antibiotic treatment in patients for whom a positive CoNS does not represent a CSB.