Colored nets have been recognized as a powerful modelling paradigm for the validation and evaluation of systems, both in terms of compact representation and aggregate state space generation. In this paper we discuss the issue of adding compositionality to a class of stochastic colored nets named Stochastic Well-formed Nets, in order to increase modularity and reuse of the modelling efforts. This requires the notion of Parametric Stochastic Well-formed net: nets in which a certain amount of information is left unspecified, and is instantiated only upon model composition. The choice of the compositional rule has been based on previous work on layered models for integrated hardware and software systems (the processes, services and resources methodology), and an example of layered modelling with Parametric Stochastic Well-formed net is presented to show the efficacy of the proposed formalism.