Since the early days of terrestrial photovoltaics, a common perception has been that ®rst generation' silicon wafer-based solar cells eventually would be replaced by à second generation' of lower cost thin-®lm technology, probably also involving a different semiconductor. Historically, cadmium sulphide, amorphous silicon, copper indium diselenide, cadmium telluride and now thin-®lm polycrystalline silicon have been regarded as key thin-®lm candidates. Any mature solar cell technology seems likely to evolve to the stage where costs are dominated by those of the constituent materials, be they silicon wafers or glass sheet. It is argued, therefore, that photovoltaics is likely to evolve, in its most mature form, to a`third generation' of high-ef®ciency thin-®lm technology. By high ef®ciency, what is meant is energy conversion values double or triple the 15±20% range presently targeted, closer to the thermodynamic limit of 93%. Tandem cells are the best known of such high-ef®ciency approaches, where ef®ciency can be increased merely by adding more cells of different bandgap to a cell stack, at the expense of increased complexity and spectral sensitivity. However, a range of other more`paralleled' approaches offer similar ef®ciency to an in®nite stack of tandem cells. These options are reviewed together with possible approaches for practical implementation, likely to become more feasible with the evolution of materials technology over the next two decades.