The SIDER database is a resource for drug and adverse drug reaction (ADR) data. It was developed to help researchers understand the mechanisms of drug action and the causes of adverse reactions. The database contains information on 1430 drugs, 5880 ADRs, and 140,064 drug–ADR pairs, representing a 40% increase over the previous version. The data is derived from package inserts and clinical trial reports, and includes frequency information for ADRs and drug indications. The database also includes a set of drug indications extracted from package inserts using natural language processing (NLP), which helps reduce false positives by identifying terms that do not correspond to ADRs. The SIDER database was created in 2010 when no such resource was freely available for academic researchers. It has since been used in many studies, including identifying metabolic dysregulation as a cause for ADRs, investigating the effect of essential proteins on ADRs, and predicting drug indications. SIDER has also been used as a benchmarking set for text-mining methods that extract ADR data from the literature. Other databases have incorporated data from SIDER, such as ADReCS, which combined SIDER data with an independent annotation effort and added an ontology of ADRs. The database is available at http://sideeffects.embl.de/. It includes a website that allows users to search for drugs and ADRs, and to trace drug–side effect pairs to the drug labels. The complete data set of side effects and the data set of indications are available for download in text format, including PubChem and MedDRA identifiers. The database is also available on a GitHub repository where users can contribute errors they detect in SIDER. The authors can then remove the source of the errors in future versions of SIDER.The SIDER database is a resource for drug and adverse drug reaction (ADR) data. It was developed to help researchers understand the mechanisms of drug action and the causes of adverse reactions. The database contains information on 1430 drugs, 5880 ADRs, and 140,064 drug–ADR pairs, representing a 40% increase over the previous version. The data is derived from package inserts and clinical trial reports, and includes frequency information for ADRs and drug indications. The database also includes a set of drug indications extracted from package inserts using natural language processing (NLP), which helps reduce false positives by identifying terms that do not correspond to ADRs. The SIDER database was created in 2010 when no such resource was freely available for academic researchers. It has since been used in many studies, including identifying metabolic dysregulation as a cause for ADRs, investigating the effect of essential proteins on ADRs, and predicting drug indications. SIDER has also been used as a benchmarking set for text-mining methods that extract ADR data from the literature. Other databases have incorporated data from SIDER, such as ADReCS, which combined SIDER data with an independent annotation effort and added an ontology of ADRs. The database is available at http://sideeffects.embl.de/. It includes a website that allows users to search for drugs and ADRs, and to trace drug–side effect pairs to the drug labels. The complete data set of side effects and the data set of indications are available for download in text format, including PubChem and MedDRA identifiers. The database is also available on a GitHub repository where users can contribute errors they detect in SIDER. The authors can then remove the source of the errors in future versions of SIDER.