In Western countries, vascular dementia (VaD) is the most common form of cognitive deterioration after Alzheimer's disease. Therapeutic trials in VaD have so far failed to yield satisfactory results. One explanation of this failure may be the etiological and clinical heterogeneity of the included patients. Patients with subcortical VaD, defined on a clinical and radiological basis, may constitute a more homogeneous group. Thus, we conducted a post-hoc subgroup analysis of the Scandinavian Multi-Infarct Dementia Trial that evaluated the efficacy and safety of oral nimodipine administered for 6 months in 259 patients. The original patients sample was divided on the basis of head CT in those with subcortical VaD (n=92, 45 nimodipine, 47 placebo) and those with multi-infarct dementia (n=167, 83 nimodipine, 84 placebo). While in the total trial population a treatment effect could not be proved, in this subgroup analysis, the subcortical VaD patients treated with nimodipine performed better on the majority of neuropsychological tests and functional scales in comparison with patients on placebo. No trend could be evidenced in the multi-infarct dementia patients. Treatment efficacy was in particular suggested for the Zahlen-Verbindungs-Test, Fuld-Object-Memory Evaluation, Word Fluency, and for the Instrumental Activities of Daily Living scale. The results did not reach statistical significance in this small sample. Our study preliminarily indicates that nimodipine could be effective in patients with small vessel subcortical VaD and supports the rationale for a further controlled and adequately powered trial to test nimodipine in patients with subcortical VaD.