IEEE 802.11 network deployments are ubiquitous and provide connectivity to millions of users. Interference that originates from other technologies, like simple Radio Frequency (RF) equipment, severely degrades the performance of those networks. To effectively manage wireless networks, the interference needs to be modeled and predicted. The current state of the art models are insufficient to model performance correctly. In this letter, we describe the interference as an interrupted Poisson process and use a decomposition approach to predict the latency of an interfered client from the latency of a non-interfered client. This novel approach allows for fast and easy prediction of latency in an interfered network. The results show that our method gets as close as 6% of the real value.