Sir: A patient presented to us having experienced a few days of massive epistaxis and hyposphagma (severe hemorrhage into the conjunctiva) while taking paroxetine. Pharmaceutical use was also notable for the peripheral vasodilator limaprost alfadex. To our knowledge, no such case has been reported in the literature. We believe the combination of paroxetine and limaprost alfadex led to the bleed. Case report. Ms. A, a 66-year-old woman with blood counts and blood coagulation tests within normal limits, presented in June 2005 to the orthopedic surgery department with complaints of coldness of limbs, continuing lower-extremity pain, insomnia, and loss of appetite. She underwent a nerve-root block for lumbar spinal canal stenosis and started limaprost alfadex 15 µg/day. Limaprost alfadex, an oral prostaglandin E1 derivative, is frequently used in Japan for the treatment of peripheral circulatory disturbances. Because the symptoms overall did not improve, she was referred to the psychiatry service (July 2005). She had lost 10 kg in 2 months and complained strongly of anxiety and frustration due to inability to do housework and to memory loss. She was admitted in September 2005. The patient was diagnosed with DSM-IV major depressive disorder with psychotic features based on hypochondriacal and persecutory delusions and aggression. When she refused food and medication due to delusions of being poisoned, electroconvulsive therapy (ECT) was introduced (October–November 2005, total of 11 treatments). ECT was effective, and she began to take medication again after the sixth ECT treatment. She was restarted on limaprost alfadex treatment (15µg/day). When ECT was completed in early November, paroxetine 20 mg/day was started, increasing to 40 mg/day after 2 days and to 50 mg/day after 4 days. The patient's mood remained normal. On the 29th and 30th days after starting paroxetine, she experienced massive epistaxis (more than 500 mL per episode). On the 30th day of treatment, paroxetine was decreased to 40 mg/day, but similar epistaxis recurred the next day. An otolaryngologist determined that hemorrhage was from the right Kiesselbach plexus; however, no other abnormality was observed. Complete blood count and blood coagulation findings were within normal limits. On the 36th day of paroxetine treatment, limaprost alfadex was reduced from 15 µg/day to 10 µg/day. No further epistaxis episodes occurred. Upper (day 45 of paroxetine treatment) and lower (day 50 of paroxetine treatment) gastrointestinal endoscopy were performed, and no hemorrhage was observed. Because of intermittent complaints of lower extremity coldness and pain, the dosage of limaprost alfadex was returned to 15 µg/day. Since January 2006, prescriptions for paroxetine 40 mg/day and limaprost alfadex 15 µg/day have continued, and the patient has been followed at the outpatient clinic. There has been no recurrent epistaxis as of September 2006. However, right eye hyposphagma was recently observed at a psychiatry clinic appointment. If this continues, we will consider decreasing paroxetine to 30 mg/day. Mild epistaxis during selective serotonin reuptake inhibitor (SSRI) treatment is a well-known rare adverse effect, but our patient's epistaxis was severe. Hyposphagma with paroxetine might be very rare.1 Since only a few case reports2 have suggested SSRI use alone can trigger even mild bleeding, including ecchymoses, purpura, and epistaxis, we cannot make a strong conclusion. After carefully examining the drugs in concurrent use for our patient, it is quite possible that limaprost alfadex, in combination with paroxetine, contributed to severe epistaxis and hyposphagma. When patients present with severe symptoms, including psychotic features, many clinicians use relatively high doses of antidepressants for continuation and maintenance treatment.3 Because vasodilating prostaglandins such as limaprost alfadex are used frequently in fields such as orthopedic surgery, it is important that clinicians know there may be a tendency for hemorrhage when an SSRI is also used; physicians should consider decreasing the SSRI maintenance dose in such instances.