The clinical pharmacodynamics of temazepam were investigated in patients who received spinal anaesthesia. Total plasma and cerebrospinal fluid temazepam concentrations were measured and correlated with the clinical effects. Sedation was measured by three separate methods. None, including an aggregated score of all three measures, was correlated closely with either the plasma or the cerebrospinal fluid levels (p = 0.86 and 0.12 respectively). Anxiety was measured before and after premedication. The two scores were correlated but the change in anxiety after premedication did not correlate with either the plasma or the cerebrospinal fluid concentrations (p = 0.11 and 0.45 respectively). Short-term memory was measured before and after premedication. The decline in short-term memory ability was moderately well correlated with both the plasma and the cerebrospinal fluid levels (p = 0.0005 and 0.013 respectively). With temazepam, the variation in sedative and anxiolytic effects between subjects is explained not by differences in pharmacokinetics but rather by differences in the pharmacodynamic response. Because sedative and anxiolytic effects are poorly correlated, but the amnesic effect is well correlated with temazepam concentrations, different sites of action for these effects are suggested.