This paper addresses the problem of optimal route selection in wireless mesh networks (WMNs) where intra-flow and inter-flow interferences have a negative impact on the network performance. Several routing metrics have been proposed in the literature to find paths that minimize interferences and thus maximizes throughput. However, most of these metrics consider either inter-flow interferences or intra-flow interferences and a few consider both types of interferences. In this paper, we propose an efficient new routing metric, called MBP (Metric Based on Probabilities), that aims to choose routes with high throughput, low intra-flow and inter-flow interferences between a source and a destination. MBP is based on the Minimum Loss (ML) metric and a well known physical interference model. Simulation results show that MBP outperforms some existing metrics, namely hop count, WCETT and iAWARE, in terms of throughput, delay and packet loss.