The purpose of the present study was to analyse specific advantages of calculated parameter images and their limitations using an optimized echo-planar imaging (EPI) technique with high spatial and temporal resolution. Dynamic susceptibility contrast magnetic resonance imaging (DSC-MRI) was performed in 12 patients with cerebrovascular disease and in 13 patients with brain tumours. For MR imaging of cerebral perfusion an EPI sequence was developed which provides a temporal resolution of 0.68 s for three slices with a 128 x 128 image matrix. To evaluate DSC-MRI, the following parameter images were calculated pixelwise: (1) Maximum signal reduction (MSR); (2) maximum signal difference (delta SR); (3) time-to-peak (Tp); and (4) integral of signal-intensity-time curve until Tp (SInt). The MSR maps were superior in the detection of acute infarctions and delta SR maps in the delineation of vasogenic brain oedema. The time-to-peak (Tp) maps seemed to be highly sensitive in the detection of poststenotic malperfused brain areas (sensitivity 90%). Hyperperfused areas of brain tumours were detectable down to a diameter of 1 cm with high sensitivity (> 90%). Distinct clinical and neuroradiological conditions revealed different suitabilities for the parameter images. The time-to-peak (Tp) maps may be an important advantage in the detection of post-stenotic "areas at risk", due to an improved temporal resolution using an EPI technique. With regard to spatial resolution, a matrix size of 128 x 128 is sufficient for all clinical conditions. According to our results, a further increase in matrix size would not improve the spatial resolution in DSC-MRI, since the degree of the vascularization of lesions and the susceptibility effect itself seem to be the limiting factors.