Membrane phospholipid abnormalities in the brain neurons may be implicated in the pathophysiology of neuropsychiatric disorders. In the absence of methods to directly examine the levels of brain membrane phospholipids in vivo in human subjects, peripheral cells and platelets have been used as models in this field. We previously reported a method to determine the relative amounts of eight individual platelet membrane phospholipid classes using two-dimensional thin-layer chromatography, and scanning-laser densitometry (Mallinger et al., 1993). Here we report the test/retest reproducibility of these platelet membrane phospholipid measures in healthy human subjects (n = 12) who were studied on two different occasions separated by a 3-week interval. The mean intra-subject coefficients of variation were 3.1-18%, and the intra-class correlation coefficients (ICCs) were 0.41-0.68. These findings are consistent with a low to moderate variability, and moderate reliability of these individual platelet membrane phospholipid measures over the period studied. When this method is applied for longitudinal studies of psychiatric populations, the degree of variability has to be considered in the interpretation of the results.