In the research presented here a multipath learning curve model of the transition of Czochralski silicon solar cell modules from a satellite application product to a terrestrial application product is developed. The validity of this model is tested by applying the historical data for the initial period of this transition ( i.e. 1974 – 1979). After the success of the multipath learning curve model in predicting the price dynamics for 1974–1979 is demonstrated, forecasts are made for 1980–1985. The results are a set of forecast prices for each year, corresponding to estimated confidence levels of 10%, 50% and 90%. The forecasts for 1985 (in 1979 U.S. dollars) are U.S. $2.2 W p −1 (at 10% estimated confidence level), U.S. $3.1–$3.2 W p −1 (at 50% estimated confidence level) and U.S. $4.8 W p −1 (at 90% estimated confidence level). On the basis of the solar cell module price forecasts, and the associated market value forecasts, the near-term (1985) outlook for industrial investment can be summarized as follows. 1. (1) The 1985 Czochralski silicon solar cell module business does not represent an attractive investment for the major companies involved in silicon materials or device and integrated circuit manufacture. Significant resource commitment cannot be reasonably expected from this sector until the late 1980s, if at all. 2. (2) The 1985 solar cell module business represents a very attractive m market for the small companies which are at present in this market. The intrinsic resource base of such companies is too small, however, to support the required investment level over the 1980–1985 period. To participate in this growth industry, they must acquire financial backing from venture capital or major energy corporations. Recent developments indicate that several of the international energy corporations have decided to make the major investments that are required to develop the Czochralski silicon solar module technology base. If an alternative route of development based on silicon ribbon, sheet or web (100 μm thick) technology is taken, then there is an expectation of about a fivefold reduction in cost per square meter for the silicon material. However, a strong commitment to this technology is realistic only if the solar cells produced from this material have an efficiency competitive with Czochralski-wafer-based cells, or if the cell-processing and/or module assembly costs can be substantially reduced relative to the Czochralski technology. A primary conclusion is that the form ( e.g. wafer versus ribbon) of silicon crystal material chosen for emphasis is less important than the research and development emphasis that must be placed on reducing the “core” cost element, the high purity polycrystalline silicon feedstock. The historical data inputs to this study represent, by necessity, the reality of an evolving photovoltaics manufacturing technology that has not yet shown significant impacts from the research and development program sponsored by the U.S. Department of Energy. The learning curve methodology is most appropriate to this situation, i.e. the development of a product through an adaptive and evolutionary manufacturing process scenario. The impact of revolutionary process changes and product designs on a broad front, when they occur, will terminate the evolutionary development sequence for which the learning curve methodology is most appropriate. Therefore, there is no fundamental conflict between this learning curve analysis and studies in which a more revolutionary manufacturing scenario is assumed.