Homocysteic acid (HA) has been suggested as a pathogen in a mouse model of Alzheimer's disease (AD), 3xTg-AD. However, it is not established whether HA is involved in humans. We investigated the relationship between urinary HA levels and Mini-Mental State Examination (MMSE) scores in AD patients (n = 70) and non-AD controls (n = 34). We found a positive, statistically significant relationship between the two variables (the urinary HA level and MMSE score) (r = 0.31, p = 0.0008, n = 70). This relationship was stronger in females than males (r = 0.43, p = 0.005, n = 44 in females; r = 0.48, p = 0.02, n = 22 in males). The urinary HA levels were significantly different in AD patients than controls (AD: 8.7 ± 7.5, n = 70; non-dementia control: 13.3 ± 9.4, n = 34, p < 0.01). In addition, aging and smoking were found as lowering factors for urinary HA levels. Our preliminary study showed a negative, statistically significant relationship between blood HA (micromole) and urine HA levels (r = −0.6, p = 0.007, n = 19), and between blood HA levels and MMSE scores (r = −0.79, p = 0.0000518, n = 19). On the basis of these results, we speculate that reduced urinary excretion induces elevated HA levels in blood, resulting in cognitive dysfunctions. This study also suggests that HA may be a candidate of neurotoxins for uremic encephalopathy. Since amyloid-β increases HA toxicity and HA is an agonist of N-methyl-D-aspartic acid (NMDA) receptor, we speculate that elevated blood HA affects the brain cognitive function through NMDA receptor-mediated toxicity in AD.