The Electric Network Frequency (ENF) signal can be captured in multimedia signals recorded in areas of electrical activity. This has led to the emergence of many forensic applications based on the use of ENF signals such as validating the time-of-recording of an ENF-containing multimedia signal or estimating its recording location based on concurrent reference signals from power grids. In this paper, we examine a novel application based on the use of the ENF signal that seeks to estimate the power grid in which an ENF-containing multimedia signal was recorded without relying on the availability of concurrent power references. We derive features based on the statistical differences in ENF variations between different grids to serve as signatures for the grid-of-recording of an ENF-containing signal. We use these features in a multiclass machine learning system that is able to identify the grid-of-recording of a signal with a high accuracy.