We build on a previous statistical model for distributed systems and f
ormulate it in a way that the deterministic and stochastic processes w
ithin the system are clearly separable. We show how internal fluctuati
ons can be analysed in a systematic way using Van Kanpen's expansion m
ethod for Markov processes. We present some results for both stationar
y and time-dependent states. Our approach allows the effect of fluctua
tions to be explored, particularly in finite systems where such proces
ses assume increasing importance.