This review examines the methods used to study adverse drug reactions (ADRs) and to quantify drug risk in pharmacovigilance. Beside analysis of spontaneous reports (the basic and universal alert method in pharmacovigilance), epidemiological methods can also be used. Classical methods are based on intensive recording, case-control or cohort studies, and clinical trials. Other more recent methods have been applied to pharmacovigilance. Using recent personal data, the authors present and discuss these new methods. Case/non-case studies quantify drug risk based on a pharmacovigilance database, and are used as signals in pharmacovigilance. Analysis of laboratory data can also be used to evaluate biological ADRs such as elevated liver enzyme and creatine phosphokinase levels. Cross-linkage studies can be used to minimize the consequences of under-notification in pharmacovigilance. Finally, ADR perception analysis is a useful method to evaluate the perceived social and medical importance of drug risks in different populations (public, pharmacists, general practitioners, medical specialists).