High resolution measurement in both exercised skeletal and cardiac tissue made radially outward from capillaries and longitudinally parallel to capillaries by Gayeski and Honig (1986, 1986a,b) and Honig and Gayeski (1987) indicate shallow variation of tissue PO2 and the absence of strong causal relation between the PO2 at a point and the proximity of that point to the nearest active capillary. Proposed as a model for the analysis of this tissue PO2 distribution, so contrary to the expectations of Krogh type models, are a class of multicellular tissue cylinder models. Each cylinder is penetrated by many parallel capillaries. In order to better represent the natural irregularities of the skeletal and cardiac tissue both with regard to radial placement and the stagger of the capillary inlets, the following types of models both with and without yoglobin are examined: regular square arrays where the capillary PO2 levels are random, uniform capillary PO2 levels but random capillary positions, and those with both the capillary PO2 levels and the positions are random. The results of the model calculations show that the superposition of the oxygen diffusion fields of all the capillaries produce a tissue PO2 distribution with the properties: (1) lower tissue PO2 levels than those predicted by Krogh theory, (2) significant non-local contributions to the PO2 at a point in the tissue which greatly reduces this correlation between PO2 at a point and its proximity to an active capillary, (3) shallow transcellular PO2 variation.