Objective: The objective of this study was to predict the diagnosis of bacteraemia as a function of the time at which the automated BacT/Alert system continuously detects microorganism growth. Methods: A retrospective study of a database of 1334 patients with a positive blood culture between January 2011 and June 2013 was conducted. Together with the final blood culture results and the patient's history, growth was then analysed to assess whether it represented true bacteraemia or bacterial contamination. The earliest detection times of bacterial growth in each batch of blood cultures were analysed in a blinded fashion after classification. Results: In total, 590 batches of blood cultures corresponded to true bacteraemia and 744 to bacterial contamination. In the bacteraemia group, the median growth time was 12.72 h (interquartile range (IQR) 10.08–17.58 h). In the contaminated blood culture group, the median growth time was 20.6 h (IQR 17.04–32.16 h) (p < 0.001). Analysis of the receiver operating characteristics (ROC) curve (area under the curve 0.80, 95% confidence interval 0.771–0.826) showed that 90% of the contaminants grew after 14.7 h (sensitivity 90.5%, specificity 63.4%, positive predictive value (PPV) 65.9%, negative predictive value (NPV) 90.7%). Forty-five percent of the bacteraemia organisms grew in under 12 h (sensitivity 45.3%, specificity 95%, PPV 87.8%, NPV 68.7%). Microorganisms such as Candida sp and Bacteroides sp presented median growth times significantly longer than those of the other microorganisms. The administration of antibiotics in the week prior to bacteraemia was found to delay the growth time of microorganisms (p < 0.001). Conclusions: Knowing the time to detection of microorganism growth can help to distinguish between true bacteraemia and bacterial contamination, thus allowing more timely clinical decisions to be made, before definitive microorganism identification.