Biochemical markers for AD would be of great value both to improve the clinical diagnostic accuracy in scientific studies and to increase the knowledge of the pathogenesis of the disorder. One of the main features of AD is a degeneration of synapses. Therefore, we examined if chromogranin A (CrA), the major protein of large dense-core synaptic vesicles, in cerebrospinal fluid (CSF) may be of value as a biochemical marker for the synaptic function in AD. The mean concentration of CrA in CSF was about 7.5 times higher than its concentration in serum, and there was no significant correlation between CSF-CrA and the blood-brain barrier function (measured as the CSF/serum albumin ratio), nor between CSF-CrA and serum-CrA. These findings suggest that the major portion of CSF-CrA is locally produced within the CNS. There were no significant differences in CSF-CrA between the AD (n = 29), vascular dementia (n = 13), and age-matched control (n = 9) groups (99.9 ± 58.9 ng/ml, 108.0 ± 69.4 ng/ml, and 115.1 ± 44.4 ng/ml, respectively). However, when the AD group was subdivided into AD type I (n = 12) and AD type II (n = 17), a lower concentration of CSF-CrA was found in AD type I (72.8 ± 28.9 ng/ml) compared with controls (115.1 ± 44.4 ng/ml), p < 0.02, and compared with AD type II (119.1 ± 67.5 ng/ml), p < 0.05, while CSF-CrA did not significantly differ between AD type II and controls. These findings suggest that CSF-CrA has a potential as a biochemical marker for the synaptic degeneration in AD type I, and gives further support for the relevance of identifying the AD type I (pure AD) subgroup in scientific studies.