We consider a stationary continuous time process with a finite or countable state space, which is usually called a stationary pure jump process because it changes its state by jumps. Relationships between the stationary distributions at an arbitrary time, just before and after jump instants are obtained by using conditional sojourn times in states and in sets of states. For a skip free process, these distributions have product form expressions in terms of the conditional mean sojourn times. The results are applied to queueing models. We extend some known relationships between queueing characteristics to batch arrival and batch serve queues with stationary inputs. They also give a unified approach for truncation expressions for finite queues.