BACKGROUND This study defines the cerebrospinal fluid (CSF) pharmacokinetics of neostigmine after intrathecal injection in humans and its effect on CSF acetylcholine, and it correlates physiologic effects with neostigmine dose and CSF acetylcholine concentrations. METHODS The CSF was sampled via an indwelling spinal catheter in 12 volunteers receiving intrathecal neostigmine (50-750 microg) and analyzed for neostigmine and acetylcholine. Pharmacokinetic and pharmacodynamic analyses were performed with NONMEM. Effect-site models linked the time course of the neostigmine concentration with the time course of analgesia. RESULTS Acetylcholine concentrations increased from <20 pmol/ml at baseline to >100 pmol/ml within 15 min of neostigmine injection. The pharmacokinetics of intrathecal neostigmine were best described by a triexponential function with an absorption phase. Individual predicted concentrations varied 100-fold. Post hoc Bayesian estimates described the observed neostigmine concentrations with a median error of 22% and did not show systematic model misspecification. Individual estimates of effect site concentration producing a 50% maximal effect for foot visual analog scale analgesia correlated with the magnitude of individual CSF neostigmine concentrations. CONCLUSIONS Intrathecal neostigmine concentrations can be well described by a triexponential disposition function, but the intersubject variability is large. The correlation between intersubject variability in concentration and intersubject variability in 50% maximal effect for foot analgesia suggests that both are offset by a common scalar, possibly the distance from the site of injection to the sampling and effect sites. These data provide the basis for the hypothesis of "observation at a distance" to describe the pharmacodynamics of intrathecally administered drugs.