The deployment of both outdoor and indoor Small Cell Base Stations (SCBSs) has attracted significant interest in the wireless industry. However, a critical concern in a large scale deployment is the efficient control of the small cell’s transmission mode. In this paper we propose a load aware approach for dynamic idle mode selection where the load distribution is estimated by the “RF Fingerprints” of the users. Our approach allows pilot signals and most of the processing power of a SCBS to be completely switched off when no active user is in its vicinity, or when the required Quality of Service (QoS) can be provided by the underlying macro base stations. Such an approach significantly reduces the energy consumption of the small cells as well as reducing the pilot pollution and signalling overhead. We evaluate the efficiency of our approach using field measurements in central Dublin as well as with system-level simulations. We show that the proposed method is capable of identifying idle cells with an average prediction error rate of 1.9%. Moreover we show that it has the potential to achieve an average reduction of 90% in kWh power consumption compared to the concurrent operation of all SCBSs.