Abstract Population pharmacokinetics and simulation techniques currently play an important role in new drug development. This paper illustrates the potential value of those methods in postmarketing safety assessment, using buflomedil in elderly patients as an example. We retrospectively assessed the risk of buflomedil overdosing associated with the latest dosing recommendations of the French Drug Agency (AFSSAPS). First, buflomedil concentrations measured in 24 elderly patients were analysed with a nonparametric population approach. Then, the pharmacokinetic model was used to perform a 1000‐patient Monte Carlo simulation for the two recommended buflomedil dosage regimens. The maximum concentrations calculated after 10 days of therapy were compared with levels observed in reported cases of toxicity to assess the probability of overdosing. A three‐compartment model best fit concentration data. Population predictions showed little bias (−0.14 mg/L) and good precision (8.73 mg 2 /L 2 ). Overall results of the simulation study showed that the application of the two recommended dosage regimens of buflomedil was associated with overdosing ( C max > 10 mg/L) and potential toxicity in 2.9% of geriatric patients. In patients with mild renal impairment, who may receive the higher‐dosage regimen by therapeutic error, the probability of overdosing was 6.2%. Despite specific dosing recommendations in case of renal impairment, this study shows that the use of buflomedil could be associated with significant risk of overdosing in geriatric patients. Such results might have enhanced decision‐making when buflomedil safety was reassessed by AFSSAPS in 2006. The retrospective case of buflomedil illustrates how these methods may be valuable in postmarketing safety evaluation of potentially toxic drugs.