The popularity of wireless networks has increased in recent years and is becoming a common addition to LANs. In this paper we investigate a novel use for a wireless network based on the IEEE 802.11 standard: inferring the location of a wireless client from signal quality measures. Similar work has been limited to prototype systems that rely on nearest-neighbor techniques to infer location. In this paper, we describe Nibble, a Wi-Fi location service that uses Bayesian networks to infer the location of a device. We explain the general theory behind the system and how to use the system, along with describing our experiences at a university campus building and at a research lab. We also discuss how probabilistic modeling can be applied to a diverse range of applications that use sensor data.