Background and purpose: The aggregation of vascular risk factors (VRFs) can aggravate cognitive impairment in stroke-free patients. Metabolites in serum and cerebrospinal fluid (CSF) may irreversibly reflect early functional deterioration. This study evaluated small-molecule metabolites (<1,000 Da) in the serum and CSF of patients with different degrees of cerebrovascular burden and investigated the correlation between metabolism and cognitive performance associated with VRFs. Methods: The subjects were divided into a low-risk group (10-year stroke risk <= 5%), a middle-risk group (10-year stroke risk >5% and <15%), and a high-risk group (10 years stroke risk >= 15%) according to the Framingham stroke risk profile (FSRP) score, which was used to quantify VRFs. We assess the cognitive function of the participants. We semiquantitatively quantified the small molecules using liquid chromatography-tandem mass spectrometry (LC-MS/MS). The correlation between the small molecules and cognitive function, along with VRFs, was investigated to identify key small molecules and possible underlying metabolic pathways. Results: When the FSRP scores increased, the cognitive performances of the subjects decreased, specifically the performance regarding the tasks of immediate memory, delayed recall, and executive function. Seven metabolites (2-aminobutyric acid, Asp Asp Ser, Asp Thr Arg, Ile Cys Arg, 1-methyluric acid, 3-tert-butyladipic acid, and 5 alpha-dihydrotestosterone glucuronide) in serum and three metabolites [Asp His, 13-HOTrE(r), and 2,5-di-tert-Butylhydroquinone] in CSF were significantly increased, and one metabolite (arachidonoyl PAF C-16) in serum was significantly decreased in high-risk group subjects. Among these metabolites, 1-methyluric acid, 3-tert-butyladipic, acid and Ile Cys Arg in serum and 13-HOTrE(r), 2,5-di-tert-butylhydroquinone, and Asp His in CSF were found to be negatively related with cognitive performance in the high-risk group. Arachidonoyl PAF C-16 in serum was found to be associated with better cognitive performance. Caffeine metabolism and the tricarboxylic acid cycle (TCA cycle) were identified as key pathways. Conclusions: 1-Methyluric acid, 3-tert-butyladipic acid, arachidonoyl PAF C-16, and Ile Cys Arg in serum and 13-HOTrE(r), 2,5-di-tert-butylhydroquinone, and Asp His in CSF were identified as potential biomarkers of vascular cognitive impairment (VCI) at the early stage. Caffeine metabolism and the TCA cycle may play important roles in the pathophysiology of VRF-associated cognitive impairment.