Differential privacy is a notion of privacy originated from the community of statistical databases, and now widely adopted for the protection of confidential information in systems of various nature. We consider a probabilistic process calculus as a specification formalism for concurrent systems, and we establish a framework for reasoning about the degree of differential privacy provided by such systems. We give a compositional method to compute the conditional probabilities that relate the secret and the public information, and we investigate the constructs which do not decrease the degree of privacy under composition.