We address the problem of reliable broadcasting in spite of collisions and transmission losses. We do not aim to achieve reliable broadcast to all nodes in the network, rather, we seek to ensure reliable broadcast to a desired percentage of nodes. We first present a geometric based, probabilistic model to predict the optimal transmission range for maximizing 1-hop broadcast coverage in wireless networks as a function of range, sending rate and density. Though this adaptation technique can be incorporated into any broadcasting protocol, we extend Broadcast Protocol for Sensor Networks (BPS) [1] because of its simplicity, scalability and adaptability. The protocol can be applied under node heterogeneity in terms of transmission/ reception ranges and packet delivery rate. Through extensive analytical and experimental results, we show multi fold advantages of adapting the protocol to network conditions.