My talk will aim at giving an overview of my dissertation, which deals with probabilistic theories of causation. I will first explain that generic causation and singular causation are in dissimilar positions with regard to probabilistic theories, and I will briefly set out the questions relative to generic causation that I have tackled. Then I will focus on singular causation. Here, my starting point will be that there cannot be a satisfactory probabilistic theory of causation unless Humphreys' paradox concerning conditional propensities is solved, and a propensity interpretation for conditional probabilities is produced. The bulk of the paper will be devoted to describe the lines along which this may be done and to formulate one particular proposition. I will finally examine the consequences of this proposition as far as the idea of a probabilistic theory of singular causation is concerned.