Existing models of nutrient transfer in arbuscular mycorrhizal (AM) symbioses are inadequate as they do not explain the range of real responses seen experimentally. A computer simulation model was used to evaluate the novel hypotheses that mycorrhizal nutrient transfers were based solely on symbionts' internal needs, and that carbon and phosphorus transfers were quantitatively unlinked. To be plausible, simulated mycorrhizal plants would show a +/-50% variation in weight vs nonmycorrhizal controls, with a normal response distribution (mimicking a real data set). One plant and one arbuscular mycorrhizal fungus (AMF) growing in a soil volume were simulated, using C, P and nitrogen nutrient cycling and stoichiometry. C- and P-exchange rates were independent and could be varied at will. The model was tested at realistic nutrient concentrations and a full range of nutrient exchange rates. The model showed -20% to +55% range in mycorrhizal plant weight distributed close to normal, suggesting that the hypotheses were plausible. The model suggests that theoretical assumptions about mycorrhizas should be reassessed. The model worked only because the symbionts possessed incomplete information on their partner and environmental conditions. Conventional cost-benefit models do not work under these circumstances, but both mutualistic and parasitic interactions were successfully simulated.