本研究係以裁量性應計數字及其組成成分衡量盈餘管理，以審計人員產業別市場佔有率與產業別客戶集中度之綜合衡量捕捉審計品質，探討不同樣本公司之盈餘管理特性，以及審計品質對不同盈餘管理特性之影響。管理者對應計項目的認列通常有某種裁量權，此裁量權可用以顯露私有資訊而增進盈餘的價值攸關性，或可投機性地管理盈餘而減低盈餘為公司績效衡量的可靠性。所以探究盈餘管理行為是增加財務報告資訊性(information)或雜訊(noise)即非常重要。研究內容依盈餘管理特性區分為投機性盈餘管理與訊息發放性盈餘管理二大部分。每一研究內容則是先辨認盈餘管理本質，其次探討審計品質對盈餘管理之影響。在研究設計上，為避免實證代理變數之衡量誤差重大影響結果，乃採用實驗組公司與控制組配對公司相比較，以凸顯出前者之盈餘管理型態。 　　研究結果顯示，門檻公司從事盈餘管理以達前期盈餘門檻，且此盈餘管理不具未來現金流量持續性之涵義，係符合「隱藏當期不利績效、遞延本期不尋常盈餘至未來年度」以減低會計盈餘數字波動性之特性。就自願性盈餘預測此一資訊揭露工具而言，在盈餘預測偏屬樂觀之傾向下，仍有管理者自願揭露壞消息預期，即可知管理者之誠意，不失為一可資信賴之訊號。但自願預測盈餘增加，則不必然具有高公信力。甚至在管理者降低盈餘差距之動機下，易引起管理者操縱盈餘。此為財務報表使用者所需注意。就持股變動此一資訊揭露工具而言，在我國家族企業居多之情形下，增加持股並未代表任何特定之涵義。反而是減少持股所透露之訊息，值得投資人注意，尤須慎防其盈餘操縱行為。 　　面對增加所得之裁量性應計數字，審計品質有效阻止投機性盈餘管理。面對顯露私有資訊之盈餘管理，審計品質能夠協助管理者選擇最能彰顯未來現金流量預期之會計方法，以增強裁量性應計數字與未來現金流量變動之關聯性。在高審計品質所賦予之資訊公信力下，盈餘資訊在投資人之決策過程中有一定之重要性，故投資人發現裁量性應計數字增加（減低）盈餘價值攸關性，因而增加（減低）盈餘評價係數。然而對低審計品質所查核的財務報表，較低的資訊公信力，使盈餘未在投資人評價過程中扮演重要角色，因而盈餘管理特性亦不受投資人重視。 / This research examined characteristics of earnings management and effects of audit quality therein. Specifically, the magnitude of earnings management was proxied by the discretionary accruals, which are estimated using a cross-sectional version of the Jones model. Different from prior research, the audit quality was captured by a combined measure of the auditor's market share and client concentration. The market share measured the auditor's competence (industry specialty) and the client concentration measured the auditor's independence. Managers usually exercised discretion to recognize accruals. They could either signal private information to enhance earnings' value relevance, or opportunistically manage earnings to reduce the reliability of earnings as the performance measure of the firm. As a result, it was important to examine earnings management and to identify whether it would increase the informativeness or noise of financial reports. This study was organized according to the characteristics of earnings management: opportunistic earnings management and signaling earnings management. For both parts, the nature of earnings management was identified first, and then the effects of audit quality therein examined. To reduce the impact of measurement errors of proxy variables, the treatment group and controlled group were compared and contrasted to make earnings management pattern of the treatment group more evident. 　　The results showed that threshold firms managed earnings to sustain last year's earnings level. Discretionary accruals of the threshold firms were not significantly associated with future cash flow changes. Those discretionary accruals were used to hide the current unfavorable performance, or to defer current unusual earnings to the future, so that managers could reduce the volatility of accounting earnings. As a means of information disclosure, voluntary earnings forecasts tended to be optimistic, but we still could observe some voluntary forecasts which decreased earnings. In those situations, we might conclude that managers tried to convey credible signals. However, it was not necessarily credible in the case of voluntary forecasting which increased earnings. Managers would manipulate earnings to reduce the difference between forecasted earnings and reported earnings. In respect of the information signal of changes in directors' share holding percentages, it signaled nothing when directors increased holding percentages because most companies were family businesses in Taiwan. Nonetheless, investors should look out the possibility of earnings manipulation when directors decreased their holding percentages. 　　Audit quality could effectively deter opportunistic earnings management but not the income-decreasing discretionary accruals. Based on information signaling perspective of earnings management, auditors with higher audit quality tended to assist managers to choose accounting methods that might signal cash flow expectations. Stronger associations between discretionary accruals and future cash flows were observed for firms audited by higher quality auditors. Because higher audit quality might increase information credibility, earnings would play a more important role in the decision process of investors. When discretionary accruals increased (decreased) earnings' value relevance, earnings association coefficients would be increased (decreased). However, for the financial reports audited by lower quality auditors, earnings did not play an important role in the valuation process because of lower information credibility. Therefore, empirical results showed that investors of firms audited by lower quality auditors did not make decisions according to different characteristics of earnings management.