Between January 1, 2002, and September 30, 2005, 1,084 public companies restated previously issued financial statements. This represents 16 percent of the average number of listed companies, as compared to almost 8 percent of listed companies that made restatements between 1997 and 2001 (GAO 2006b, 4). Because they involve correcting financial statements previously filed with the U.S. Securities and Exchange Commission (SEC), restatements may reflect abuse of the flexibility that exists within generally accepted accounting principles (GAAP) and thus have "raised questions about the credibility of accounting practices and the quality of corporate financial disclosure and oversight in the United States" (GAO 2002, 1). Not surprisingly, academic research on restatements has grown in recent years. However, most attention has been focused on effects relative to restatement announcements. In contrast, this study examines the timeframe during which the errors or misapplication of GAAP originally occurred. Specifically, this study explores the relationship between earnings and returns and that between earnings and chief executive officers' (CEOs') cash compensation over the period for which earnings were initially misstated. Using data hand-collected from the original and amended documents filed with the U.S. Securities and Exchange Commission (SEC), after some minor adjustments such as excluding restatements related to purchased in-process research and development, no difference is found in the strength of the association between returns and earnings as originally reported versus as restated. However, decomposing the originally reported earnings into the restated (GAAP-compliant) component and restatement/error (GAAP-non-compliant) component suggests that market participants find both components to be value relevant when profits are originally reported but only the restated (GAAP-compliant) component to be value relevant when losses are originally reported. Similar results are found when analyzing the relation between earnings and CEOs' cash compensation. When firms initially report profits, both the GAAP-compliant and GAAP-non-compliant components are positively associated with CEOs' cash compensation. Although other explanations are possible, this result is consistent with the market "tolerating" misstatements by profitable firms. Another possible explanation points towards the contracting role of accounting earnings and shareholders' implicit sanction for deviation from GAAP. When firms initially report losses, the GAAP-compliant component is negatively associated with CEOs' cash compensation and the GAAP-non-compliant component is not significant. This suggests that the GAAP-non-compliant component does not play a role in contracting when losses are initially reported. Whether the market tolerates some misstatements from profit-making firms, or efficient contracting requires such deviations, are subjects of future research.