The use of ionized particle beams for thin film deposition offers the unique possibility to control the kinetic energy of the impinging particles in an energy range that greatly exceeds the thermal energy inherent to conventional growth methods. For this reason epitaxy may conceivably occur at substantially reduced substrate temperatures. This study focusses on film deposition from ionized cluster beams, as higher growth rates are achievable than with low-energy ion beams. Silver clusters have been produced in an adiabatic expansion in a supersonic beam apparatus. Mean cluster sizes of typically 30 atoms were measured in seeded nozzle beams, using helium as carrier gas. Such small sizes, which are inappropriate for cluster beam deposition, can be expected on the basis of a simple thermodynamic model that relates the average cluster size to the source parameters (pressure, temperature and nozzle diameter). It is shown that a rough estimate of the mean cluster size in a nozzle beam can be derived from the value of ψ = P 0 d( T b/ T 0) γ/(γ-1). The model predicts that considerably larger clusters of metals of semiconductors will only be formed in nozzle beams generated under extreme source conditions. On the contrary, clusters of 10 4-10 5 atoms are readily produced by evaporation of a metal in an inert-gas atmosphere. With a proper choice of source pressure and orifice diameter, intense cluster beams can be generated which can give deposition rates for silver up to 0.5 μm/min. Depending on the average cluster size, 5 to 25% of the clusters is ionized. By electrostatic and magnetic deflection these ionized clusters can be separated from the beam to study film growth at the desired kinetic energy. The crystallinity of the growing layers was monitored using reflection high-energy electron diffraction. Both for silver and for germanium the best results were obtained with cluster beams having a relatively small average cluster size, corresponding to supplied kinetic energies of the order of 1 eV/atom. Oriented crystalline growth of silver on silicon was observed at room temperature, whereas for germanium a substrate temperature of 230°C was required. Though rather speculative, a model is proposed that can accomodate the observations by considering the thermal effects induced by the collision of an accelerated cluster at the growth interface.