Using plastics or dyes like molecular semiconductors to make optoelectronic devices at low cost is a project in industry that is of great interest to the "Information Community" of this beginning of the 21st century. Some applications of this technology have already been realized including flexible flat-panel displays made with organic light-emitting diodes, organic transistors for active matrices, organic solar cells and sensors. In all these applications it is essential to deduce and understand the laws that determine the charge transport and the light emission in these materials. The goal of this thesis is to contribute to the theoretical models that describe these transport processes. In order to predict the behaviour of charge carriers in a given device, one can choose two methods. One can either numerically solve the master equations that describe the general behaviour of charges and currents in each part of the device, or one can simulate the detailed behaviour of each charge in the device by fixing the conditions of its movement using a procedure known as "Monte Carlo". In this thesis I addressed the two approaches. I have been interested successively in the prediction of the electric properties of multilayer light-emitting devices, considered as a whole. Then I was able to understand in detail the processes accruing at the organic-organic interfaces by using the Monte Carlo method. My last project involved the study of the channel of an organic transistor. In each one of these cases I took account of the characteristic features of small molecule organic materials in order to develop models for charge transport for the study and optimization of two important types of organic devices : organic light-emitting diodes and field-effect transistors. Amorphous organic semiconductors are mainly used for the fabrication of organic diodes. They are characterized by an energetically disordered density of states that is assumed to be Gaussian. The transport of charges in these materials occurs via hopping from one molecule to another in this density of states. The correlations between the energies of the molecular sites have important effects on transport ; the dependence of mobility on the applied electric field in particular. If the energetic disorder is due to the random orientations of the permanent dipoles of the molecules, then the correlations between the orientations of these dipoles can profoundly change the spatial configuration of the energetic disorder and consequently the charge transport. Due to the fact that organic diodes often comprise of multiple organic layers, understanding device behaviour at the organic-organic interfaces is of critical importance. Electrons and holes accumulate close to these interfaces and they are therefore more likely to recombine and emit light. The Coulomb interactions between the charge carriers and the energy disorder are at the origin of complex processes taking place at these accumulation regions. I studied these processes through a 3D multi-particle Monte Carlo simulation. The short-range Coulomb repulsion forbids the double-occupancy of the sites and shifts the energy level for injection of carriers across the interface upward, thus increasing the electric current of the diode. The long-range Coulomb interactions increase the electric field and the current at the interface although this increase is less important than suggested by 1D models. Effects related to the hopping process such as the effect of the exact form of the hopping law and backward or return hop at the interface are clearly discussed. The Monte Carlo simulations are cross-compared with an analytical model for hopping transport across the interface. This comparison made it possible to understand other aspects of the physical processes at organic-organic interface and supports the Monte Carlo simulations. The study of the electron-hole recombination showed a decrease of the cross-section with an increase of the applied electric field, but this decrease is less than what can be expected by the classical Langevin theory. In spite of their efficiency, Monte Carlo simulations are not attractive when modelling complex multi-layer diodes. The 1D models are better suited in this case. A 1D model called MOLED, which was developed at Laboratoire d'Optoélectronique des Matériaux Moléculaires (LOMM) for the simulation of the organic diodes, is used and described in detail in this thesis. We used MOLED to explore the effect of the energy correlations on the electric field dependence of the mobility. In fact we simulated the time-of-flight experiment, a common method used to determine of the carrier mobility. The crystalline organic semiconductors achieve mobility performances close or even better than amorphous silicon. Their use for the fabrication of field-effect transistors is thus a good choice. The models for charge transport in crystalline organic materials differ much from their counterparts for amorphous organic materials. Indeed band models are more appropriate to describe charge transport in crystalline organic materials. However it is very important to take into account the effects of electronic polarization in these models. I developed a method of calculation of these effects of polarization in the bulk of the organic semiconductor and I extended it to the case of the interface with a dielectric insulator, a situation encountered in the channel of the organic transistor. These calculations showed that when the polarity of the dielectric increases the inter-molecular transfer integral decreases. This is confirmed by experimental measurements showing increasingly low values of mobility when the permittivity of the insulator increases.