Background: Diffusion tensor imaging (DTI) is the imaging technique used in vivo to visualize white matter pathways. The corticospinal tract (CST) belongs to one of the most often delineated tracts preoperatively, although, the optimal DTI method has not been established yet. Considering that various regions of interests (ROIs) could be selected, the reproducibility of CST tracking among different centres is low. We aimed to select the most reliable tractography method for outlining the CST for neurosurgeons. Materials and methods: Our prospective study consisted of 32 patients (11 males, 21 females) with a brain tumour of various locations. DTI and T1-weighed image series were acquired prior to the surgery. To draw the CST, the posterior limb of the internal capsule (PLIC) and the cerebral peduncle (CP) were defined as two main ROIs. Together with these main ROIs, another four cortical endpoints were selected: the frontal lobe (FL), the supplementary motor area (SMA), the precentral gyrus (PCG) and the postcentral gyrus (POCG). Based on these ROIs, we composed ten virtual CSTs in DSI Studio. The fractional anisotropy, the mean diffusivity, the tracts' volume, the length and the number were compared between all the CSTs. The degree of the CST infiltration, tumour size, the patients' sex and age were examined. Results: Significant differences in the number of tracts and their volume was observed when the PLIC or the CP stood as a single ROI comparing with the two-ROIs method (all p<0.05). The mean CST volume was 40054U; SD ±12874 and the number of fibres was 259.3; SD ±87.3 when the PLIC was a single ROI. When the CP was a single ROI, almost a half of fibres (147.6; SD ±64.0) and half of the CST volume (26664U; SD ±10059U) was obtained (all p<0.05). There were no differences between the various CSTs in terms of fractional anisotropy, mean diffusivity, the apparent diffusion coefficient, radial diffusivity and the tract length (p>0.05). The CST was infiltrated by a growing tumour or edema in 17 of 32 patients; in these cases, the mean and apparent diffusion of the infiltrated CST was significantly higher than in uncompromised CSTs (p=0.04). CST infiltration did not alter the other analysed parameters (all p>0.05). Conclusions: A universal method of DTI of the CST was not developed. However, we found that the CP or the PLIC (with or without FL as the second ROI) should be used to outline the CST.