The application of vertical forests in the building industry is a popular development with promising advantages regarding sustainability. However, placing trees along a façade gives additional loads which often results in extra material use. This is undesired concerning building design, costs, and the environment. To mitigate these adverse effects, trees could be used as load-bearing structural elements. To increase the stability of such elements, it might be advantageous to connect the trees to each other. Such connections can naturally be created with inosculations: self-growing connections in which the bark and inner tissue of two trees merge. Applying living trees as structural elements requires a deeper understanding of both the botanical and structural behaviour of trees. Likewise, more must be known about the botanical and structural behaviour of self-growing connections. This research expands the knowledge on the topic of using living trees as structural elements, both as single tree elements and self-growing interconnected tree elements. This aim is reached by proposing a structural model of (interconnected) trees, which is verified by comparing the outcome of the model with the outcome of winching tests. Furthermore, a design is made in which living trees are used as structural elements in a vertical forest case study. This lays the groundwork on how to approach a living tree design, both from a structural and botanical point of view. The structural model is verified by carrying out winching tests on the Living Tree Pavilion, including one single tree, one pair of cross-connected trees and one pair of parallel-connected trees. Additionally, winching tests are performed on a tree in an airpot, in which boundaries constrain the root system. During the winching tests, a force is applied to the tree system, and the elongation is measured on several locations of the tree, providing insight into the strain distribution. Additionally, the displacement of the trees is measured at the height of the force application. Based on geometry measurements, the trees are modelled as solids in a finite element software. From the models of the interconnected trees, a compound solid is created which behaves like a single solid. The winching load is applied to the models to allow for a comparison between the results of the winching tests and the models. The geometry measurements show indications that leaning trees create an oval cross-section, which is influenced by the presence of inosculations. The winching tests show that an unconstrained root system is stiffer than a root system constrained by an airpot. Furthermore, the tests show that interconnected trees do not have favourable stiffness qualities compared to single trees. A comparison between the model and test results indicates that the finite element model is a plausible representation of reality for the single tree, the parallel-connected trees, and the out-of-plane results of the cross-connected trees. The finite element model fitted poorer with the in-plane winching test results of the cross-connected trees. More research is needed to determine whether diverging tree characteristics, in the direction that is rarely subjected to loads, could explain the discrepancy in measured and modelled behaviour. Two designs are created in which living trees of the Wonderwoods vertical forest carry the loads of a plant container. There are three reasons why cross-connected trees are not favourable over a design with single trees. First, the single trees can bear the plant containers at a younger age. Second, the risk of trees not creating suitable inosculations is high. Third, as interconnected trees share one container, the competition for space can become fierce. This research concludes that a system of interconnected trees as structural elements is not favourable over a system of single trees. This is mainly because no clear advantages in terms of strength and stability could be found in both the winching tests and the design for Wonderwoods.