This paper presents a novel site-specific interference analysis model for Wireless Personal Area Network. This model not only provides accuracy rather than the stochastic model, but also computationally efficient prediction compared to the Monte-Carlo (MC) simulation using the ray tracing algorithm. It can be considered when the interference issues are arisen in indoor Wireless Personal Area Network (WPAN) at power point of view. In the proposed model, the concept of mean free distance as a site-specific parameter is employed using geometric optics and geometric probability to take into account configuration of environment. As a result, a closed-form expression to obtain the average interference power and the mean free distance as a site-specific parameter at the two different indoor scenarios are derived. The proposed model has been validated by comparing Signal-to-Interference-Noise-Ratio (SINR) with those of calculated result from stochastic model and MC simulation using ray-tracing. This reveals its better ability to predict interference power than conventional stochastic model and shows faster computation speed than MC simulation.