We study the loss of spatial coherence in the extended wave function of fullerenes due to collisions with background gases. From the gradual suppression of quantum interference with increasing gas pressure we are able to support quantitatively both the predictions of decoherence theory and our picture of the interaction process. We thus explore the practical limits of matter wave interferometry at finite gas pressures and estimate the required experimental vacuum conditions for interferometry with even larger objects.PACS numbers: 03.65.Yz,39.20.+q Matter wave interferometers are based on quantum superpositions of spatially separated states of a single particle. However, as is well known, the concept of wave-particle duality does not apply to a classical object which by definition never occupies macroscopically distinct states simultaneously. By performing interference experiments with particles of increasing complexity one can therefore probe the borderline between these incompatible descriptions.It is still a matter of debate how to explain the quantum-to-classical transition in a unified framework. Some theories contain an element beyond the unitary evolution of quantum mechanics [1, 2] -which includes the 'collapse' of the wave function as taught in many standard textbooks. Decoherence theory, on the other hand, remains within the framework of the quantum theory [3,4,5]. It explains the decay of quantum coherences as being caused by the interaction of the quantum object with its environment.So far, several decoherence experiments in atom interferometry focused on the loss of coherence due to scattering of a single [6,7] or a few [8] laser photons by an atom. Other authors proposed or realized schemes to encode which-path information in internal atomic degrees of freedom, thereby reducing the interference contrast as well, in spite of a negligible change in the atomic centerof-mass state [9,10]. These studies are complemented by experiments which quantitatively followed the decoherence of a coherent photon state in a high-finesse microwave cavity [11] or of the motional state of a trapped ion [12]. However, all these experiments worked with few-level systems and engineered environments.In the present letter we quantitatively investigate a mechanism which seems to be among the most natural and most effective sources of decoherence in our macroscopic world, namely collisions with gas particles. From the controlled suppression of quantum interference as a function of the gas pressure we are able to test both the predictions of decoherence theory and our picture of the collisional interaction.We note that the effect of atomic collisions in an atom interferometer was already investigated in [13]. How- ever, decoherence effects were not observed in these experiments, since the detected atoms did not change the state of the colliding gas sufficiently to leave behind the required path information for decoherence. In contrast to that, our experiment uses massive C 70 -fullerene molecules, and is based on a Talbot-L...