Video telephony requires high-bandwidth and low-delay voice and video transmissions between geographically distributed users. It is challenging to deliver high-quality video telephony to end-consumers through the best-effort Internet. In this paper, we present our measurement study on three popular video telephony systems on the Internet: Google+, iChat, and Skype. Through a series of carefully designed active and passive measurements, we are able to unveil important information about their key design choices and performance, including application architecture, video generation and adaptation schemes, loss recovery strategies, end-to-end voice and video delays, resilience against random and bursty losses, etc. Obtained insights can be used to guide the design of applications that call for high-bandwidth and low-delay data transmissions under a wide range of "best-effort" network conditions.