The South African government is introducing a poverty-reduction policy that will supply households with a monthly 50 kWh free basic electricity (FBE) subsidy. We show that FBE distorts the energy choices of poor households by encouraging them to cook with electricity, whereas alternatives such as liquefied petroleum gas (LPG) can deliver a similar cooking service at a much lower cost to society. An alternative energy scheme, such as providing households with clean energy credits equivalent in value to the FBE's cost, could deliver additional energy services worth at least 6% of total household welfare (and probably much more) at no additional public cost; those benefits are so large that they would cover the entire cost of LPG fuel needed to implement the scheme. The analysis is extremely sensitive to the coincidence of electric cooking with peak power demand on the South African grid and to assumptions regarding how South Africa will meet its looming shortfall in peak power capacity. One danger of FBE is that actual peak coincidence and the costs of supplying peak power could be much less favorable than we assume, and such uncertainties expose the South African power system to potentially very high costs of service. (c) 2005 Elsevier Ltd. All rights reserved.