For patients suffering from bloodstream infections (BSI) molecular diagnostics from whole blood holds promise to provide fast and adequate treatment. However, this approach is hampered by the need of large blood volumes. Three methods for pathogen DNA isolation from whole blood were compared, i.e. an enzymatic method (MolYsis, 1-5 ml), the novel nonenzymatic procedure (Polaris, 1-5 ml), and a method that does not entail removal of human DNA (Triton-Tris-EDTA EasyMAG, 200 mu l). These methods were evaluated by processing blood spiked with 0-1000 CFU/ml of Staphylococcus aureus, Pseudomonas aeruginosa and Candida albicans. Downstream detection was performed with real-time PCR assays. Polaris and MolYsis processing followed by real-time PCRs enabled pathogen detection at clinically relevant concentrations of 1-10 CFU/ml blood. By increasing sample volumes, concurrent lower cycle threshold (Ct) values were obtained at clinically relevant pathogen concentrations, demonstrating the benefit of using larger blood volumes. A 100% detection rate at a concentration of 10 CFU/ml for all tested pathogens was obtained with the Polaris enrichment, whereas comparatively lower detection rates were measured for MolYsis (50-67%) and EasyMAG (58-79%). For the samples with a concentration of 1 CFU/ml Polaris resulted in most optimal detection rates of 70-75% (MolYsis 17-50% and TTE-EasyMAG 20-36%). The Polaris method was more reproducible, less labour intensive, and faster (45 minutes (including Qiagen DNA extraction) vs. 2 hours (MolYsis)). In conclusion, Polaris and MolYsis enrichment followed by DNA isolation and real-time PCR enables reliable and sensitive detection of bacteria and fungi from 5 ml blood. With Polaris results are available within 3 hours, showing potential for improved BSI diagnostics.