The proliferation of wireless and mobile devices has fostered the demand for context-aware applications, in which location is one of the most significant contexts. Multilateration, as a basic building block of localization, however, has not yet overcome the challenges of 1) poor ranging measurements; 2) dynamic and noisy environments; and 3) fluctuations in wireless communications. Hence, multilateration-based approaches often suffer from poor accuracy and can hardly be employed in practical applications. In this study, we propose Quality of Trilateration (QoT) that quantifies the geometric relationship of objects and ranging noises. Based on QoT, we design a confidence-based iterative localization scheme, in which nodes dynamically select trilaterations with the highest quality for location computation. To validate this design, a prototype network based on wireless sensor motes is deployed and the results show that QoT well represents trilateration accuracy, and the proposed scheme significantly improves localization accuracy.