In order to characterize ADP-ATP and creatine-creatine phosphate (Cr-CrP) shuttles a minimal mathematical model with two compartments and cyclic turnover of matter was designed. The 'mitochondrial' compartment contained 'ATP-synthase' and 'mitochondrial ereatine kinase' (mitCK). The 'cytoplasmic' compartment consisted of 'ATPase', 'cytoplasmic creatine kinase' (cytCK) and an 'ADP-binding structure'. The exchange of metabolites between these compartments was limited. Different levels of cytCK and mitCK expression as welt as different exchange rate constants between the compartments were assigned to obtain several different modes. Every steady state obtained in the presence of low ATPase activity ('resting' conditions) was then disturbed by a steep activation of ATPase ('muscle performance') and afterwards the transition to a new steady state was followed in time. The ATP-buffering capacity of the system initially acquired by cytCK expression significantly increased after additional mitCK supplement. Nevertheless, even the complete Cr-CrP shuttle failed to maintain a high [ATP]/[ADP] ratio during long term 'muscle performance' due to the rate limiting CK-transphosphorylation in the mitochondria. The facilitated diffusion of Cr and CrP was not critical, and the model worked with the same efficiency even at equal permeabilities for nucleotides and guanidines. Under 'resting conditions' the main flux of matter went through the Cr-CrP shuttle, resulting in 'pumping' of CrP. This ensured a 40 s delay in the [ATP] decrease at 'work'. The partial systems without mitCK were not as effective, and this delay was 0-10 s. However, the ADP-ATP shuttle was of more importance at the steady state achieved under ' working' conditions.