In this paper we present an integration of PLinda, a probabilistic extension of Linda, and StoKLAIM, a stochastic extension of KLAIM. In the resulting language, StoPKLAIM, the execution time of coordination primitives is modeled by means of exponentially distributed random variables, as in StoKLAIM, the choice of the primitive to be executed among conflicting ones is thus resolved by the race condition principle, and the choice of the tuple to be retrieved by a single input/read operation in case of multiple matching tuples is governed by the weight-based probabilistic access policy of PLinda. The language represents a natural development and integration of previous results of the SENSORIA Project in the area of probabilistic and time-stochastic extensions of Tuple Space based coordination languages. The formal operational semantics of StoPKLAIM is presented and an example of modeling is provided.