The development of new medicines suffers from attrition, especially in the development pipeline. Eight out of nine drug candidates entering the clinical testing phase fail, mostly due to poor safety and efficacy. The low predictive value of animal models, used in earlier phases of drug development, for effects in humans poses a major problem. In particular, drug disposition can markedly differentiate in experimental animals versus humans. Meanwhile, classic in vitro methods can be used but these models lack the complexity to mimic holistic physiological processes occurring in the human body, especially organ–organ interactions. Therefore, better predictive methods to investigate drug disposition in the preclinical phase are needed, for which recent developments in multiorgan-on-chip methods are very promising. To be able to capture human physiology as good as possible, multiorgan-on-chips should feature 1) human cells endogenously expressing main transporters and metabolizing enzymes; 2) organ models relevant for exposure route; 3) individual organs-on-chip connected in a physiologically relevant manner; 4) a tight cellular barrier between the compartments; 5) organ models properly polarized in 3D; 6) allow for sampling in all major compartments; 7) constructed from materials that do not absorb or adsorb the compound of interest; 8) cells should grow in absence of fetal calf serum and Matrigel; 9) validated with a panel of compounds with known characteristics in humans; 10) an integrated computer model translating concentrations to the human situation. Here, an overview of available systems is presented and the difficult route towards a fully validated system is discussed.