The authors present the results of the utilization of a pharmacokinetic prediction test for lithium posology. Based on a single point (plasma lithium determination 24 h after a single dose) such a test aims to adapt the posology as soon as the second day of treatment rather than after one week as clinicians must wait for a steady state to be achieved. Built on the previous work of Perry, the test targeted the plasma lithium level at 0.8 +/- 0.1 mmol l(-1). Thirty-one patients took part in the study. There were two drop-out cases and the results were available for 29 patients: among them, 51% had their plasma level in the targeted zone. Although there was no control group, the prediction test often allowed us to use a higher dose than the usual fixed dose whose amount is limited by the risk of overdosing for the slower metabolizers.