Antibody-drug conjugates (ADCs) are one of the fastest growing classes of oncology therapeutics. After half a century of research, the approvals of brentuximab vedotin (in 2011) and trastuzumab emtansine (in 2013) have paved the way for ongoing clinical trials that are evaluating more than 60 further ADC candidates. The limited success of first-generation ADCs (developed in the early 2000s) informed strategies to bring second-generation ADCs to the market, which have higher levels of cytotoxic drug conjugation, lower levels of naked antibodies and more-stable linkers between the drug and the antibody. Furthermore, lessons learned during the past decade are now being used in the development of third-generation ADCs. In this Review, we discuss strategies to select the best target antigens as well as suitable cytotoxic drugs; the design of optimized linkers; the discovery of bioorthogonal conjugation chemistries; and toxicity issues. The selection and engineering of antibodies for site-specific drug conjugation, which will result in higher homogeneity and increased stability, as well as the quest for new conjugation chemistries and mechanisms of action, are priorities in ADC research.