Pharmacokinetic and pharmacodynamic responses to caffeine (2.5 mg/kg) were compared between ten healthy self-rated poor sleepers and ten normal sleepers. Sleep pattern assessed by the Pittsburgh Sleep Quality Index (PSQI). There was no significant difference in mean estimated daily caffeine consumption between the groups. The poor sleepers had significantly higher scores for neuroticism on the Eysenck Personality Questionnaire (EPQ) and anxiety on the Hospital Anxiety Depression (HAD) scale, compared with normal sleepers. Caffeine pharmacokinetics were assessed by measurement of saliva caffeine concentrations. Poor sleepers showed significantly greater variability in caffeine C_max, clearance and half-life, compared to normal sleepers. Pharmacodynamic measures included heart rate, blood pressure, visual analogue scales for concentration, vigilance and relaxation, psychomotor performance [Digit Symbol Substitution Test (DSST) and tapping rate (TR)] and EEG activity [Contingent negative variation (CNV), auditory evoked potential and power spectral analysis]. Prior to caffeine administration, poor sleepers compared to normal sleepers had faster heart rates, lower ratings for concentration and relaxation, poorer performance on the DSST, greater CNV magnitude, faster peak alpha frequency and lower delta, theta and beta power. These differences persisted after caffeine ingestion and overall differences between the groups on these measures were significant (P<0.01–0.001). Post-dose, but not pre-dose, scores for vigilance and TR were significantly lower overall in poor compared with normal sleepers. Despite the baseline differences between poor and normal sleepers, the changes following caffeine administration were similar in direction and magnitude in both groups.