For estimation of regional myocardial blood flow with O-15 water PET, a few modifications considering partial volume effect based on single compartment model have been proposed. In this study, we attempted to quantify the degree of heterogeneity and to show the effect of tissue flow heterogeneity on partition coefficient(λ) and to find the relation between perfusable tissue index(PTI) and λ by computer simulation using two modified models. We simulated tissue curves for the regions with homogeous and heterogenous blood flow over a various flow range(0.2-4.0ml/g/min).Simulated heterogeneous tissue composed of 4 subregions of the same or different size of block which have different homogeneous flow and different degree of slope of distribution of blood flow. We measured the index representing heterogeneity of distribution of blood flow for each heterogeneous tissue by the constitution heterogeneity(CH). For model I, we assumed that tissue recovery coefficient (FMME) was the product of partial volume effect(FMMF) and PTI. Using model I, PTI, flow, and FMM were estimated. For model II, we assumed that partition coefficient was another variable which could represent tissue characteristics of heterogeneity of flow distribution. Using model II PTI, flow and λ were estimated. For the simulated tissue with homogeneous flow, both models gave exactly the same estimates of three parameters. For the simulated tissue with heterogeneous flow distribution, in model I, flow and FMM were correctly estimated as CH was increased moderately. In model II, flow and λ were decreased curvi-linearly as CH was increased. The degree of underestimation of λ obtained using model II, was correlated with CH. The degree of underestimation of flow was dependent on the degree of underestimation of λ. PTI was somewhat overestimated and did not change according to CH. We conclude that estimated λ reflect the degree of tissue heterogeneity of flow distribution. We could use the degree of underestimation of λ to find the characteristic heterogeneity of tissue flow and use λ to recover the underestimated flow.