INTRODUCTION: Despite a narrow therapeutic index, lithium remains a cornerstone for the treatment of bipolar disease. As lithium poisoning may result in life-threatening neurotoxicity, measurement of the lithium concentration is mandatory in drug monitoring as well as for the diagnosis and prognostic evaluation of lithium poisoning. However, toxic symptoms do not always correlate with plasma concentrations. Therefore, more reliable indicators have been proposed, including measurement of the red blood cell (RBC) lithium concentration and the RBC-to-plasma lithium ratio. PLASMA AND RBC LITHIUM CONCENTRATIONS: Few studies have reviewed the relative utility of these measurements both in monitoring therapy and in poisoning, and they have involved only small numbers of subjects. Moreover, factors influencing plasma and RBC lithium concentrations are numerous, including gender, age, dosage, treatment duration, co-medications, and underlying diseases. In treated patients, investigated using in vivo magnetic resonance spectroscopy, there is a significant correlation between plasma and brain lithium concentrations in steady-state conditions. In contrast, lithium transport across erythrocytes markedly differs from its transport into the central nervous system, questioning the relevance of measuring the RBC lithium concentration. In poisoned patients, plasma and RBC lithium concentrations follow a parallel decline irrespective of the type of poisoning. CONCLUSIONS: Based on present evidence, measurement of the RBC lithium concentration and the calculation of the RBC-to-plasma lithium ratio offer no important clinical advantage over the measurement of the plasma lithium concentration, which remains the most important variable to monitor in lithium-treated or lithium-poisoned patients.