THis STUDY ATTEMPTS to specify and test a general model of fixed investment that contains, as special cases, marginal-efficiency-of-investment, and hypotheses. It applies this model to annual data covering the period 1947 to 1965 and 45 steel firms in the United States. Parameters are estimated by regressions using pooled cross-section, pooled time-series, yearly cross-section, crosssection time-series, and firm time-series structurings of individual-firm data, to test the relative explanatory power of each hypothesis and to investigate various subsidiary but important issues. The general model suggests that simultaneous interaction of demand and supply schedules of physical and financial investment determines the optimum rate of investment and rental price of capital. A reduced-form equation for optimal investment states that investment is a distributed-lag function of past changes in net investment and of current and past changes in adjusted sales, internal funds, the rate of interest, debt saturation, and the ratio of wages to rental price. Under different assumptions, the capacity-accelerator, marginal-efficiency-of-invest-' ment, and liquidity hypotheses are each obtained from the general model. The capacity-accelerator relation includes past changes in net investment and current and past adjusted changes in sales. The marginal-efficiency-of-investment relation contains past changes in net investment and current and past changes in the supply price of investment, profits, the interest rate, and the market value of the firm. The liquidity relation comprises past changes in net investment and current and past changes in the supply price of investment, profits, internal funds, the interest rate, and debt saturation. Pooled cross-section regressions for 1951-1965 and separate regressions for each year in this period give better results, by standard statistical criteria, for the general hypothesis. Findings from time-series by firm, pooled time-series, and cross-section time-series regressions are similar. The performance of individual variables points to adjusted changes in sales as the most consistent determinant of investment. Other variables in the general hypothesis are found to be important at different times and for different firms. However, no systematic connection between the significance of individual variables and stages of business fluctuations is discerned. Changes in the ratio of wages to rental price add little to the explanatory power of the general model regardless of the definition used for rental price. Cross-section results for variants of the general model and the marginal-efficiency-of-investment and liquidity hypotheses indicate that changes in profits serve primarily as a proxy for changes in internal funds, and to a lesser degree for adjusted changes in sales. A final important result is the rejection of a null hypothesis that extraneous influences have affected the level or timing of investment in the steel industry during the postwar period. Years involving strikes, wars, and heavy foreign competition are associated with significant disturbances in the investment relation.