This paper describes a Multilayered Integrated Numerical Model of Surface Physics – Growing Plants Interaction (MINoSGI), which represents interactions between the dynamics of forest ecosystems and microclimate. Aiming at a large-scale study in the future, we describe forest dynamics by using area-averaged prognostic equations for thedistributions of plant density and plant weight with respect to plant height classes, instead of individual-based treatments for small-scale forest patches. Growth and mortality of plants are modelled based on the carbon balance of each plant height class. The area-averaged microclimate (e.g., light, wind speed, temperature, humidity, CO2 concentration) within the forest canopy is simulated by a one-dimensional multilayer canopy model, which includes most of the physical and physiological processes that control the forest microclimate. Owing to its multilayered framework, a direct specification is possible for the difference in the growing environment among plants of different size and species. Given hourly meteorological conditions, the model outputs energy, water, CO2 and momentum fluxes to and from a forest, of which the structure changes through competition among plants. The model's performance was tested by comparing its outputs with observed data on the development of plant size distribution taken over a 5-year period in an evergreen coniferous (Cryptomeria japonica) forest. The model produced realistic estimates of the total biomass increments during the period. The ratio of net primary production to gross primary production (=0.45) was consistent with previous estimates for temperate forests. The bimodal seasonal pattern in net ecosystem production was similar to the seasonal trend in the CO2 flux measured over a forest of the same species. Although some limitations due to the one-dimensional representation of microclimate were noticeable, the model adequately simulated distributions of annual growth rate, plant weight and diameter across plant height classes. Since the basic equations can be extended to include the effect of spatial variability with marginal increase of computational costs, the present model framework is feasible for large-scale studies.