In the last decade, large-scale privatization programs have transformed the structure of corporate ownership in many countries, and are a major cause of the development of their capital markets. They have contributed to nondebt financing of the public deficit, attracted foreign capital and technology, and promoted the return of flight capital. Finally, the large task of privatization of state-owned enterprises in Eastern Europe has proven to be one of the most difficult in the transition to a market economy. This paper examines the history of large-scale privatization plans in several countries, and offers some revealing insight on the strategy used by governments to achieve large sales on the capital markets. Our results are relevant for investment bankers involved in advising governments on privatization sales, for institutions interested in assessing the risk of portfolio investment in emerging economies, and for potential joint-venture investors in former state-owned enterprises. We document remarkable similarities across different sale programs. Firms are typically sold gradually, even when control is transferred rapidly. Sales of utilities tend to be more gradual than for manufacturing firms in competitive markets. The form of the sale is usually a fixed-price offer, which is often heavily underpriced. We compare a popular interpretation of gradual sales, namely limited market capacity, with a confidence-building rationale. Partial sales are usually seen as a necessity dictated by the size of privatization programs relative to the existing public equity market. The argument is that rapid sales may swamp the market and depress sale prices. However, capital markets are by definition forward-looking; thus, their current size is a misleading indicator of their potential capacity, particularly as privatization tends to take place alongside a general reform process which emphasizes markets and restores confidence in the domestic economy. However, a reform process may be subsequently reversed under pressure by entrenched interest groups; thus investment in a privatized firm is exposed to the fisk of future adverse policy changes. A privatizing government must therefore overcome investors' diffidence about its regulatory and trade policy after the sale. In order to maximize proceeds, a government may retain a stake in the firm for some time (provided managerial control is transferred immediately). This willingness to bear part of the Policy fisk will lead to higher confidence and sale prices. The progress of sales may accelerate overtime as the policy credibility of reform becomes established. Similarly, early sales may be deliberately underpriced in order to convince the market to absorb larger sales. Although the two explanations have similar empirical implications, we identify several features of sale programs which appear to be consistent with the reputation-building hypothesis rather than with the notion of temporary market capacity constraints. The risk of adverse policy changes is likely to be particularly serious for firms with large fixed investment programs which operate as natural monopolies, such as utilities, or for protected manufacturers. The evidence indicates that sales of such policy-sensitive firms tend to be more gradual and exhibit larger underpricing. In the literature on initial public offerings (IPOs), a very gradual sale structure with large underpricing is optimal when the seller has superior information over the value of the firm. However, a government is not likely to know more than private investors about asset values. Moreover, most of these firms are large, well-known utilities with stable profitability and limited competition. We conclude that the main source of uncertainty in these cases concerns future regulatory policy. In contrast, complete sales are usually associated with manufacturing firms in competitive markets. Moreover, underpricing is, on average, greater in privatization sales (and particularly so for oligopolistic firms) than in the sale of private firms. The market capacity view does not distinguish between a sequence of complete sales and a program of staggered partial sales of comparable size. The evidence suggests that governments prefer to sell stakes in several firms simultaneously as opposed to sequencing complete sales, presumably building credibility over different industries at the same time. Moreover, a common feature of all programs is that the time profile of privatization proceeds tend to increase rapidly. This is inconsistent with a fixed capacity constraint, but can be rationalized as indicating a gradual buildup in investors' confidence. Finally, the fact that larger initial sales become more common over time suggests that as confidence increases, more rapid sales become feasible. Our conclusion is that the data gathered provide evidence in favor of the confidence-building hypothesis. It is consistent both with the traditional interpretations of the structure of initial public offerings of private firms and with the specific risks associated with privatization sales.