Objective: Pyrithyldione, a sedative-hypnotic drug with a poor clinical pharmacological development, was associated with anecdotal cases of agranulocytosis in the 1940s in the USA, in the 1960s and 1970s in the ex- Democratic Republic of Germany and in the 1980s in Japan. We describe the estimation of the risk of agranulocytosis associated with its use in Spain, which led to its withdrawal from the market. Methods: In collaboration with the haematology units of all the hospitals in a defined area (3.3-3.9 x 106 inhabitants), all cases of agranulocytosis meeting strict diagnostic criteria were identified. Each case - defined as an episode of agranulocytosis - was reviewed by a haematologist without knowledge of previous drug exposures. Cases and age-, gender- and hospital-matched controls were interviewed with a structured questionnaire about previous drug exposures. In addition, in order to estimate the risk of pyrithyldione-associated agranulocytosis through a case-population approach, its consumption among the cases was compared with its consumption among the general population. Results: After a follow-up of 66.5 x 106 person-years, 330 cases of agranulocytosis (230 community cases) were assembled. Reliable information on previous exposures was obtained for 204 cases. They were compared with 1314 controls. Eleven patients (14 cases, 6.9%) and zero controls had been exposed to pyrithyldione. The adjusted OR was 200.11 (CI 95% 22.62-∞). All patients were female; none had a fatal outcome; three exhibited positive rechallenge; and all had concomitantly taken other drugs. Although pyrithyldione was a prescriptiononly medicine, only 8% had been dispensed with medical prescriptions. Assuming the worst case, i.e. that all the exposed cases could be attributed to pyrithyldione, the incidence was 35.6 cases per 100,000 patient-years (95% CI, 18.9-60.9), which gives a risk ratio estimate of 109.6 (57.5-191.5) if compared with the incidence of agranulocytosis among the non-exposed population [3.26 cases (CI 95% 2.83-3.71) per 106 inhabitants and per year]. Discussion: Pyrithyldione was viewed by pharmacists as a mild hypnotic, and apparently this had conferred to this drug an unjustified image of safety The National Commission of Pharmacovigilance recommended to the Ministry of Health its withdrawal from the market when eight cases of agranulocytosis had been identified. However, it took more than 2 years to withdraw it, and six additional cases occurred in the study area. This illustrates the need for quick regulatory action when pharmacoepidemiological data suggest an unfavourable benefit/risk ratio.