The aim of this thesis is to study multicomponent adsorption equilibria in the liquid phase. Most experimental data on adsorption reported in the literature are for binary systems with only a handful of studies on ternary systems. This is mainly because multicomponent adsorption data are considerably more difficult to measure than pure component data. As a consequence, theories that can predict multicomponent adsorption equilibria using only pure component data are frequently used even though they often fail even for binary systems. For screening purposes however, such theories are important. When screening for appropriate adsorbents, a minimum amount of experimental work is desirable. In chapter 2 of this thesis, we propose a methodology that allows the calculation of multicomponent adsorption equilibria using limited experimental data. A six-component system comprising traces of butanal, 2-ethyl-2-hexenal, 2,6-dimethyl-cyclohexanone, 2,4,6-trimethylphenol and 2,4,6-trimethylanisole in liquid toluene was chosen as a test system. The compounds were chosen such that they are volatile, organic, and have different molecular sizes and functional groups. Pure component adsorption isotherms in the sodium form of zeolite Y (NaY) were computed using Monte Carlo simulations. The Ideal Adsorbed Solution Theory (IAST) was used to compute the multicomponent adsorption equilibria using the pure component data obtained from molecular simulations. To calibrate the model, three binary experimental data points were used. We show that the combined molecular simulation - IAST approach can be used for this six-component system to predict the adsorption behavior in NaY reasonably well. To improve the accuracy predictions of IAST, activity coefficients are used to describe the non-ideal behavior of the adsorbed phase. This approach is called the Real Adsorbed Solution Theory (RAST). However, to this date, there are no predictive models available to describe the activity coefficients for the adsorbed phase. As an approximation, models taken from the gas-liquid phase equilibria are used. Models for activity coefficients such as Wilson and NRTL are commonly used to describe the adsorbed phase activity coefficients. These activity coefficient models depend on temperature and composition, while the pressure dependence is usually neglected. The adsorbed phase activity coefficients are dependent on temperature, composition and a third thermodynamic variable called the spreading pressure. To account for this difference, either the spreading pressure is considered constant or further approximations are used. In chapter 3 of this thesis, molecular simulations are used to study a 2D-lattice model to generate activity coefficient data at constant spreading pressure. The obtained data is used to check the accuracy of the Wilson and NRTL models for evaluating adsorbed phase activity coefficient data. We show that the Wilson and NRTL models cannot describe the adsorbed phase activity coefficients for slightly non-ideal to strong non-ideal mixtures. A new methodology is introduced for predicting adsorption of mixtures based on a simple 2D-lattice model coupled with the segregated sites approach. The segregated model assumes that the competition for adsorption occurs at isolated adsorption sites. Molecules from different adsorption sites cannot interact with each other and both adsorption sites are in contact with the bulk phase. The segregated 2D-lattice model provides accurate predictions for the system CO2-N2 in the hypothetical zeolite PCOD8200029, but fails in predicting the adsorption behavior of CO2-C3H8 in all-silica MOR-type zeolite. The predictions of the segregated IAST model are superior to those of the 2D-lattice model. The multicomponent Langmuir model, the multicomponent dual-site Langmuir model, IAST, and segregated IAST models are commonly used in dynamic models for adsorption processes, even though they usually fail in predicting the behavior of multicomponent adsorption systems. In chapter 4, we investigate the accuracy of these models in predicting the adsorption behavior of a six-component mixture comprising butanal, 2-ethyl-2-hexenal, 2,6-dimethylcyclohexanone, 2,4,6-trimethylphenol, 2,4,6-trimethylanisole and toluene in in the ammonium form of Y zeolite. The multicomponent dual-site Langmuir model and the segregated IAST model work best for this system and are further used in a multicomponent breakthrough model. We show that the breakthrough model, together with the multicomponent dual-site Langmuir model (used to calculate the equilibrium isotherms), can provide a rough qualitative estimation of the breakthrough behavior for this system. The system used as case study in this thesis comprised molecules of different size, shape, functional groups and polarity. Most pure component data required for describing the multicomponent adsorption behavior of this system were estimated. Despite these challenges, the force field developed in chapter 2 was used, together with IAST, to predict the adsorption behavior of the multicomponent system with sufficient accuracy for screening purposes. In chapter 3, we showed that the commonly used activity coefficient models taken from the gas-liquid theory cannot describe non-idealities in the adsorbed phase. In chapter 4, we introduced a simple breakthrough model and investigate several equilibrium models for predicting the adsorption of the six-component mixture in the ammonium form of Y zeolite. For this system, the multicomponent dual-site Langmuir model works best. The simple breakthrough model, together with the multicomponent dual-site Langmuir model, can only provide a rough estimate of the multicomponent breakthrough curves.