BACKGROUND: Adverse drug reactions (ADRs) represent a major public health concern, with death as the ultimate adverse drug outcome. Despite the relevance of this, the frequency of fatal ADRs (FADRs) is to a large extent unknown. Although spontaneous reporting data cannot give an exact estimate of the magnitude of drug-related mortality, it may highlight the importance and large dimensions of this public health problem. OBJECTIVE: To describe the types and pattern of reported FADRs by analysing data from the national spontaneous reporting system in Italy. METHODS: The Italian Medicines Agency (AIFA) runs a pharmacovigilance database where all the individual case safety reports (since January 2001) are stored. We selected and then analysed in detail all the case reports (to the end of December 2006) in which death was reported as the outcome. We included in the study only FADR case reports with a probable or possible causality assessment, according to the criteria established by the WHO. In line with the Italian reporting form, we divided FADR reports into two groups: (i) suspected ADRs that caused death; and (ii) suspected ADRs that contributed to death. RESULTS: In the AIFA database 38 507 suspected ADR case reports were collected, of which 641 (1.66%) had a fatal outcome. We analysed 450 case reports (1.17% of total reports), 159 (35.33%) of them causing the patient's death and 291 (64.67%) contributing to death. The annual percentage of FADR reports followed a constant trend during the 6-year period. The majority of fatal reports (79%) were sent by hospital doctors. In total, 222 different drugs were suspected as causes of FADRs. 'Systemic anti-infective drugs' was the drug category associated with the highest percentage of FADRs (21.9%), followed by antineoplastic and immunomodulating agents (18.8%), and then by nervous system drugs (14.8%). Other drug categories involved in the fatal case reports were antithrombotic agents, NSAIDs and contrast media. CONCLUSIONS: The drugs most frequently involved in FADRs were drugs of wide usage with a narrow therapeutic range or those that caused serious skin or systemic allergic reactions. Ceftriaxone, ticlopidine and nimesulide were associated with the highest number of fatal case reports; the related FADRs were already known and recognized for each of these drugs. We highlight some cases reflecting probable inappropriate drug use by Italian physicians. This suggests a need for continued clinical pharmacology training and that many FADRs might be preventable through better medical and prescribing practice.