After reports from Japan of neuropsychiatric adverse events (NPAEs) in children taking oseltamivir phosphate (hereafter referred to as oseltamivir [Tamiflu; F. Hoffmann-La Roche Ltd, Basel, Switzerland]) during and after the 2004--5 influenza season, Roche explored possible reasons for the increase in reporting rate and presented regular updates to the US FDA and other regulatory authorities. This review summarizes the results of a comprehensive assessment of the company's own preclinical and clinical studies, post-marketing spontaneous adverse event reporting, epidemiological investigations utilizing health claims and medical records databases and an extensive review of the literature, with the aim of answering the following questions: (i) what the types and rates of neuropsychiatric abnormalities reported in patients with influenza are, and whether these differ in patients who have received oseltamivir compared with those who have not; (ii) what levels of oseltamivir and its active metabolite, oseltamivir carboxylate are achieved in the CNS; (iii) whether oseltamivir and oseltamivir carboxylate have pharmacological activity in the CNS; and (iv) whether there are genetic differences between Japanese and Caucasian patients that result in different levels of oseltamivir and/or oseltamivir carboxylate in the CNS, differences in their metabolism or differences in their pharmacological activity in the CNS. In total, 3051 spontaneous reports of NPAEs were received by Roche, involving 2466 patients who received oseltamivir between 1999 and 15 September 2007; 2772 (90.9%) events originated from Japan, 190 (6.2%) from the US and 89 (2.9%) from other countries. During this period, oseltamivir was prescribed to around 48 million people worldwide. Crude NPAE reporting rates (per 1,000,000 prescriptions) in children (aged < or =16 years) and adults, respectively, were 99 and 28 events in Japan and 19 and 8 in the US. NPAEs were more commonly reported in children (2218 events in 1808 children aged < or =16 years vs 833 in 658 adults) and generally occurred within 48 hours of the onset of influenza illness and initiation of treatment. After categorizing the reported events according to International Classification of Diseases (9th edition) codes, abnormal behaviour (1160 events, 38.0%) and delusions/perceptual disturbances (661 events, 21.7%) were the largest categories of events, and delirium or delirium-like events (as defined by the American Psychiatric Association) were very common in most categories. No difference in NPAE reporting rates between oseltamivir and placebo was found in phase III treatment studies (0.5% vs 0.6%). Analyses of US healthcare claims databases showed the risk of NPAEs in oseltamivir-treated patients (n = 159,386) was no higher than those not receiving antivirals (n = 159,386). Analysis of medical records in the UK General Practice Research Database showed that the adjusted relative risk of NPAEs in influenza patients was significantly higher (1.75-fold) than in the general population. Based on literature reports, NPAEs in Japanese and Taiwanese children with influenza have occurred before the initiation of oseltamivir treatment; events were also similar to those occurring after the initiation of oseltamivir therapy. No clinically relevant differences in plasma pharmacokinetics of oseltamivir and its active metabolite oseltamivir carboxylate were noted between Japanese and Caucasian adults or children. Penetration into the CNS of both oseltamivir and oseltamivir carboxylate was low in Japanese and Caucasian adults (cerebrospinal fluid/plasma maximum concentration and area under the plasma concentration-time curve ratios of approximately 0.03), and the capacity for converting oseltamivir to oseltamivir carboxylate in rat and human brains was low. In animal autoradiography and pharmacokinetic studies, brain : plasma radioactivity ratios were generally 20% or lower. Animal studies showed no specific CNS/behavioural effects after administration of doses corresponding to > or =100 times the clinical dose. Oseltamivir or oseltamivir carboxylate did not interact with human neuraminidases or with 155 known molecular targets in radioligand binding and functional assays. A review of the information published to date on functional variations of genes relevant to oseltamivir pharmacokinetics and pharmacodynamics and simulated gene knock-out scenarios did not identify any plausible genetic explanations for the observed NPAEs. The available data do not suggest that the incidence of NPAEs in influenza patients receiving oseltamivir is higher than in those who do not, and no mechanism by which oseltamivir or oseltamivir carboxylate could cause or worsen such events could be identified.