Abstract: The influence of assay variation and duration of storage on changes in cerebrospinal fluid (CSF) levels of tau and phosphorylated (P)-tau with time was evaluated in 112 patients with various neurological disorders.: These patients (aged 66±9 years, 52% male), referred to our memory clinic, underwent two spinal taps (mean interval 19 months) and the baseline samples were assayed twice in a sandwich enzyme-linked immunosorbent assay (ELISA): once after the first spinal tap (A1) and once in a separately stored aliquot (A2) simultaneous with the follow-up sample (B).: Coefficients of variances (CVs) of tau and P-tau levels determined in repeated spinal taps (ΔB–A2) measured in one assay (10.9% and 7.6%) were lower (p<0.01) than the CVs observed in two different (ΔB–A1) assays (16.5% and 11.7%). The CVs of tau and P-tau measurement of one CSF sample repeated on two occasions (ΔA1–A2) were 12.3% and 8.6%. A difference in mean P-tau level was found if the same CSF samples were repeatedly measured in two different ELISAs (A1–A2).: Longitudinal CSF tau and P-tau are best measured in one assay resulting in a lower variability compared to measurement in two different assays. The within person variability in levels of these markers currently limits the use of these ELISAs in a longitudinal clinical setting.Clin Chem Lab Med 2008;46:1300–4.