A double blind crossover trial of baclofen against placebo in elderly stroke patients was discontinued because the drug produced an unacceptably high level of drowsiness. In a subsequent study baclofen 10 mg was given orally to 12 elderly stroke patients, and drug concentrations measured from a series of plasma samples. A group of healthy subjects given the same dose in a previous study were used as controls. Elderly patients took longer to achieve peak plasma baclofen concentrations, but healthy controls had higher peak values and eliminated the drug more rapidly; areas under the curve were similar in the two groups. Simulations based on mean data suggest that increased drowsiness in the elderly was probably not due to changes in the drug's pharmacokinetic behaviour.