Forests provide numerous crucial ecosystem services to society that are sensitive to climate and to extreme climate events such as droughts. One of the services of interest is the ability to sequester the greenhouse gas carbon dioxide (CO2). Species mixture and structural heterogeneity has been suggested to reduce the effect of climate variability on forests. It is important to understand how increasing diversity and heterogeneity will help dampen the effect of climate events on CO2 uptake. Currently, knowledge about the ability of different forests to resist or recover from the effect of climate events is limited. To mitigate this knowledge gap, this thesis examines the implications of forest structure on CO2 fluxes. Chapter 1 provides the general background of the topic. Chapter 2 examines the implications of structural diversity for seasonal and annual CO2 fluxes in two temperate deciduous forests for a period of 11 years. The two forest sites have similar mean stand age and near-identical climate conditions but different stand structure. The main question asked was how management and related structural diversity may affect CO2 fluxes. We show that the annual net ecosystem productivity (NEP) was on average 13 % higher in the managed, even-aged, and homogenous forest, than in the unmanaged, uneven-aged, and structurally diverse forest. The homogenous forest was observed to have, however, stronger sensitivities of seasonal NEP and gross primary productivity (GPP) to environmental variables. Chapter 3 relies on data from 21 Fluxnet sites to explore the effect of nine structural parameters on the temporal stability of light-saturated photosynthetic capacity (GPP1000) and on its resistance to changes in water availability during droughts. The study addresses two questions, (a) Do structurally diverse forests have lower variation in annual GPP1000? (b) Are structurally diverse forests more resistant to drought events? The results show that unmanaged forests and forests managed as high forests, which have higher basal areas and tend to be older and more diverse in size than coppice forests, had more stable annual GPP1000. The differences between individual sites in anomalies in GPP1000 in response to droughts were mostly explained by growing season air temperature. Forest structure could have influenced the response to droughts, but in our case the structure effect could not be separated from environmental effects. Chapter 4 presents a new model of soil water and related fluxes in forests, Forest Soil Water Model - FSWM, developed in the R environment. The model is suitable for predicting soil water in a wide range of forest soils. FSWM incorporates the Gash model for interception, the Ritchie model for soil evaporation and the Richards equation for soil water movement. FSWM’s performance was evaluated against soil water measurements at 12 sites. The model performance was good for deciduous broadleaf forests, moderate for mixed forests and evergreen needle leaf forests. FSWM offers flexibility in simulating soil horizons with different depths and it is helpful when comparing modelled with observed values at different soil depths. With these characteristics, FSWM is a flexible and freely available tool for ecosystem and hydrological research. Additionally, two co-author papers are included in the appendix. The first paper assessed the net ecosystem CO2 exchange (NEE), total evapotranspiration and net primary production of two neighbouring beech (Fagus sylvatica L.) forests in central Germany differing in site management. We found the interannual variability was higher in the managed, even-aged stand, and the unmanaged forest was a weaker sink of CO2 during a dry year. The second paper investigated the factors influencing the interannual variability (IAV) of photosynthetic capacity at light saturation, a key ecosystem functional property determining gross primary productivity. The study found that the older and species rich forest had reduced IAV of GPP1000. In general, the results of this thesis support the idea that unmanaged forests, mostly older and diverse, have lower interannual variability in NEP, GPP and GPP1000 as the result of their adaptation to the habitat by selecting appropriate species, developing structure to make best use of the light, water, and nutrient resources. During droughts, the effect of the forest structure was not clear. More research covering a large range of different sites is still required to get definitive results involving more structural attributes and sites from different climates.