In further study of red blood cell (RBC) and plasma choline concentrations in 160 elderly subjects, we found no significant differences in RBC/plasma choline ratios among depressed, demented, and healthy subjects. On the basis of a smaller sample, we had earlier reported that a significantly higher proportion of Alzheimer patients had RBC/plasma choline ratios greater than 1.9. Thus, it now appears that static RBC choline levels cannot be recommended as a specific marker of Alzheimer's dementia. However, within subgroups of these diagnostic categories, determined by RBC/plasma choline ratios less than or equal to 1.9 or greater than 1.9, consistent differences in electroencephalographic (EEG) sleep measures were found. The subgroup of demented patients with a RBC/plasma choline ratio greater than 1.9 was more impaired on the Blessed Dementia Rating Scale and had less rapid eye movement (REM) sleep than the subgroup with a choline ratio less than or equal to 1.9. Similarly, depressives with a choline ratio less than or equal to 1.9 had a lower REM latency than depressives with a choline ratio greater than 1.9. Finally, depressed-demented (i.e., mixed-symptom) patients with a choline ratio greater than 1.9 showed less sleep continuity disturbance but more indeterminate non-REM sleep (reflecting loss of spindles and K-complexes) than those with lower choline ratios. These differences parallel those previously reported for diagnostically "pure" depressed and demented patients, and they suggest a possible link between peripheral RBC/plasma choline measures and central nervous system function as reflected in sleep physiological alterations.