BACKGROUND: In many neurological diseases, intracranial pressure (ICP) is elevated and needs to be actively managed. ICP is typically measured with an invasive transducer, which carries risks. Non-invasive techniques for monitoring ICP (nICP) have been developed. The aim of this study was to compare three different methods of transcranial Doppler (TCD) assessment of nICP in an animal model of acute intracranial hypertension. METHODS: In 28 rabbits, ICP was increased to 70-80 mmHg by infusion of Hartmann's solution into the lumbar subarachnoid space. Doppler flow velocity in the basilar artery was recorded. nICP was assessed through three different methods: Gosling's pulsatility index PI (gPI), Aaslid's method (AaICP), and a method based on diastolic blood flow velocity (FVdICP). RESULTS: We found a significant correlation between nICP and ICP when all infusion experiments were combined (FVdICP: r = 0.77, AaICP: r = 0.53, gPI: r = 0.54). The ability to distinguish between raised and 'normal' values of ICP was greatest for FVdICP (AUC 0.90 at ICP >40 mmHg). When infusion experiments were considered independently, FVdICP demonstrated again the strongest correlation between changes in ICP and changes in nICP (mean r = 0.85). CONCLUSIONS: TCD-based methods of nICP monitoring are better at detecting changes of ICP occurring in time, rather than absolute prediction of ICP as a number. Of the studied methods of nICP, the method based on FVd is best to discriminate between raised and 'normal' ICP and to monitor relative changes of ICP.