Abstract Real-time quaking-induced conversion (RT-QuIC) allows the amplification of miniscule amounts of scrapie prion protein (PrP^Sc). Recent studies applied the RT-QuIC methodology to cerebrospinal fluid (CSF) for diagnosing human prion diseases. However, to date, there has not been a formal multi-centre assessment of the reproducibility, validity and stability of RT-QuIC in this context, an indispensable step for establishment as a diagnostic test in clinical practice. In the present study, we analysed CSF from 110 prion disease patients and 400 control patients using the RT-QuIC method under various conditions. In addition, “blinded” ring trials between different participating sites were performed to estimate reproducibility. Using the previously established cut-off of 10,000 relative fluorescence units (rfu), we obtained a sensitivity of 85 % and a specificity of 99 %. The multi-centre inter-laboratory reproducibility of RT-QuIC revealed a Fleiss’ kappa value of 0.83 (95 % CI: 0.40–1.00) indicating an almost perfect agreement. Moreover, we investigated the impact of short-term CSF storage at different temperatures, long-term storage, repeated freezing and thawing cycles and the contamination of CSF with blood on the RT-QuIC seeding response. Our data indicated that the PrP^Sc seed in CSF is stable to any type of storage condition but sensitive to contaminations with blood (>1250 erythrocytes/μL), which results in a false negative RT-QuIC response. Fresh blood-contaminated samples (3 days) can be rescued by removal of erythrocytes. The present study underlines the reproducibility and high stability of RT-QuIC across various CSF storage conditions with a remarkable sensitivity and specificity, suggesting RT-QuIC as an innovative and robust diagnostic method.