THE process of selecting individual securities, and particularly common stocks, for investment portfolios has several dimensions. But while the basic structure of the portfolio selection process is fairly well established, considerable controversy seems to exist as to the means by which these dimensions should be implemented. It is our purpose to review critically a few major areas of conflict currently extant in investment analysis. Hopefully, this might enable practicing analysts to identify more clearly the issues involved and the possible implications of the alternative points of view. For example, it is generally agreed that the process should be oriented to and conditioned by specific portfolio objectives, such as growth, income requirements, quality limitations, and diversification targets. But as these matters are essentially in the domain of overall portfolio strategy rather than security analysis per se, they are regarded as outside the scope of this review. However, it might be noted that considerable controversy currently exists in this area as to the usefulness of computer programs to suggest and delineate the characteristics of efficient or optimum portfolios. It is my tentative opinion that existing models for portfolio structures are likely to be of very limited practical use as they seem to involve some highly questionable assumptions in order to obtain required quantifications, such as that the entire spectrum of risk factors can be adequately measured by the relative price volatility of individual issues. A second crucial dimension of security selection is concerned with appraising the potential economic values (primarily prospective earnings and dividend flows and the risks surrounding their achievement) embodied in the industries and related companies that appear eligible to meet the portfolio requirements. The content of most of the literature, as well as of our series of C.F.A. examinations, suggests that mastery of this broad area is considered to be of prime importance to the professional development of financial analysts and portfolio managers. And rightly so. For while the random walk theory of security prices may be convincing with respect to the problem of obtaining short-term trading results, this theory does not conflict with the proposition that long-term investment returns will largely be a function of the long-term economic performance of particular industries and companies. Although the concept that investment results are likely to be heavily related to corporate performance in a long-term sense is generally accepted, some recent contributions to the field have alleged that the implementation methodology should be completely revolutionized. For example, Lerner and Carleton allege that a critical investigation of the past financial statements to reveal potential problems of consistency and comparability of reported income and balance sheet data can be largely discarded because accounting and disclosure standards have improved to the point where the underlying data require no critical review.' Moreover, they allege that financial risk factors no longer require appraisal because of the greatly improved stability features of the economy; in lieu thereof, they suggest elegant mathematical techniques to develop the theoretical effects of assumed patterns of various management decisions and economic data on security values.2 However, the empirical evidence would suggest that these allegations are seriously in error. Largely because of recurrent tinkering with the tax rules for national economic policy reasons, a strong case can be