A tensegrity is a lightweight space reticulated structure consisting of compression members — struts — surrounded by a network of tension members — cables — that provide rigidity and stability. They can be easily dismantled and therefore, they provide innovative possibilities for reusable and modular structures. To date, tensegrity construction has been limited to sculptures. The number of full-scale prototypes built is increasing though few have been tested experimentally statically. Tensegrities are flexible structures and are thus often governed by serviceability criteria. They are able to adapt their shape by changing their selfstress, and when equipped with sensors and actuators, they can actively adapt to changing environments. In this way, they have the potential to become a part of an exhibition rather than merely provide shelter for one. Until now, most structural control research in civil engineering has focused on active control of structures in order to enhance safety under extreme loading. While maintaining serviceability was mentioned in early work as a goal of structural control, there has been little investigation in this area. Improving tensegrity performance througth active control needs to meet the following challenges: construction and design, structural analysis, finding and applying control commands. Full-scale prototypes of three-module, five-module and unique active fivemodule, modular and reusable tensegrity structures have been built and successfully tested. Shape is adapted through changing the length of a limited number of bars. Each module contains six struts and twenty-four cables of three different lengths. Good joint design is a priority for this type of structure. The thesis gives a detailed description of an assembly process, pin-joint design, elements and control set-up. Polyester reinforced fiber glass bars lighten the structure. Tensegrities are non-linear, highly coupled structures, that are sensitive to asymmetric loads and small environmental changes. Numerical and experimental tests show that active control satisfy serviceability criteria with a limited number of actuators. Tests show that the structure behaves linearly when subjected to vertical loads applied to a single joint. Non-linearities are detected for small displacements for loads applied to several joints and for adjusting combinations of telescoping compression members. Therefore, simplifications such as load superposition are not possible. Simulation of the nonlinear behavior using dynamic relaxation, an explicit analysis method, proved successful for predicting the response. Behavior and the control objectives, such as maintaining a constant roof slope, do not have closed loop form solutions. Therefore, stochastic search algorithms are required to find control commands. Simulated annealing search and PGSL proved successful for determining control commands. Storing good control commands improves control efficiency. Quasi-static control, through elongating active struts one by one, is effective and safe for maintaining the slope. The present work contributes to developing structures that, through computational control and recording previous good adjustments, improve their performance during service life. Finally, an extended active control concept is suggested for self repair.