This work was supported by the MRSEC program of the National Science Foundation under Award Number DMR-0079996. An experimentally-obtained percolation threshold for high-angle random grain boundary networks in alloy 304 stainless steel is compared to thresholds predicted by percolation theory. A discrepancy occurs in the two values (0.46 experimental and 0.65 theoretical). Possible reasons for the discrepancy are explored. The grain boundary network appears to be composed of two distinct sub-networks, with the 'outer' network acting as the dominant contributor to the percolating paths.