Background Neurovascular coupling (NVC) can be assessed using transcranial Doppler (TCD) measured task-activation of cerebral blood flow velocity (CBFv). However, not all individuals show consistent responses. The aim of this study was to develop a robust, objective, method to identify non-responders to task-activation. New Method Using five-minute seated resting (non-stimulated), bilateral CBFv data from 135 healthy participants, the cross-correlation function peak (CCF) between the population coherent average and each individual was obtained for a randomly selected segment of data (40 s) for both hemispheres (n = 270). The variance ratio (VR) was calculated by comparing the variance in CBFv data pre- and post-random mark. The 90th percentile for non-stimulated data was used to determine the upper confidence limit of normal variation in the CCF peak value (0.53), and VR (2.59). These criteria were then applied to task-activated CBFv from 69 healthy participants for five cognitive tasks (attention, verbal fluency, language, visuospatial, memory). Results Data were accepted as responders if either CCF ≥ 0.53 or VR ≥ 2.59. The number of cases accepted as responders for each task were as follows: attention, 54–59 (78–86 %); verbal fluency, 42–48 (60–70 %); language, 51–53 (74–77 %); visuospatial, 54 (78 %); memory, 40–47 (58–68 %). Comparison with existing method Currently, there are no objective criteria for the identification of non-responders in studies of NVC. This is a new method to objectively classify non-responders to task-activation. Conclusions Using a large sample of resting CBFv data, we have set objective criteria to differentiate between responders and non-responders in task activation protocols.