Collaborative, virtual environments (CVEs) leverage desktop virtual reality technologies to enable multiple, simultaneous participants to interact and work together. As immersive, multimedia-oriented systems, CVEs typically represent participants as avatars and operate in a variety of areas including, but not limited to: entertainment, commerce, education, and research. Success of endeavors within these areas requires privacy and integrity for both participant activities and the objects upon which participants act. Security mechanisms, however, are often limited in scope and functionality, or lacking entirely. In response, this dissertation investigates and develops controls to protect the privacy and integrity of activities and objects within CVEs. We begin with a survey of 16 technologies that may be used to build a CVE. We then examine two of these technologies further, Croquet and Pro ject Wonderland, giving consideration to their respective architectures, strengths, and weaknesses. Continuing with Wonderland, we design and implement a full discretionary access control solution called WonderDAC. Finally, we carry out a realistic demonstration of Wonderland/WonderDAC with a small number of human participants.