Four different methods to measure acetylcholine (ACh) and choline (Ch) concentration, i.e. thermospray/mass spectroscopy (TS/MS), high pressure liquid chromatography/mass spectroscopy (HPLC/MS), high pressure liquid chromatography/electrochemical detection (HPLC/ECD) and gas chromatography/mass spectroscopy (GC/MS), both latter methods coupled to a solid phase extraction system were compared for their applicability to human lumbar cerebrospinal fluid (CSF). Furthermore, samples from 15 control persons and 11 patients with dementia of Alzheimer-type (DAT) were compared to search for an ACh deficit in the CSF in DAT. GC/MS was the most sensitive, but most laborious method, and HPLC/ECD was acceptably sensitive, reliable and more specific. TS/MS was not specific enough for CSF extracts and HPLC/MS was more specific, but far less sensitive. Thus, only GC/MS and HPLC/ECD were qualified to detect ACh in human CSF extracts. Comparison of GC/MS and HPLC/ECD revealed highly correlated levels of ACh (r = 0.999). Using HPLC/ECD, ACh concentrations were greatly reduced in the DAT group (3.75 +/- 1.40 pmol/ml CSF) as compared to the controls (6.14 +/- 1.39 pmol/ml CSF), but the difference between controls and DAT patients was not statistically significant due to the number of samples below detection limit (8 out of 11 samples in DAT, 7 out of 15 in controls). Ch concentrations were not statistically significant between the two groups. The data show that methodological limitations preclude a widespread clinical application of determining ACh in the human CSF. Despite of reductions of ACh in the CSF in DAT, the determination of Ach in the CSF is not suitable for diagnostic purposes.