This thesis examines factors which may explain content variation in sell-side analysts’ reports. There are two main objectives: (i) to ascertain whether the extent of accounting information contained in these reports varies with firm and analysts’ characteristics; and (ii) to examine whether the tone and readability of the reports vary with analysts’ incentives to produce optimistic research. Based on a sample of 288 reports on 144 S&P 500 firms, the first objective was addressed using a manual content analysis to examine accounting themes, while the second objective was addressed using automated content analysis based on context-specific and user-defined wordlists. The empirical results indicate that the extent of use of accounting information in analysts’ reports varies across firm characteristics but such variation only partly reflects its relevance for valuation as suggested by the value relevance literature. Moreover, analysts’ incentives are influential as reports issued by analysts employed by investment banking firms or those in possession of the Chartered Financial Analysts qualification contain more references to forward-looking accounting information. Patterns of strategic reporting are also identified, as analysts employed by investment banking firms issue less readable reports compared to analysts employed by independent research firms following the same company. Further, readability is lower when the reports are less optimistic, indicating a tendency to obfuscate bad news through more complex reporting. Overall the findings are consistent with an impression management perspective as it reveals that content of analysts’ reports may not be entirely objective but influenced by analysts’ incentives to promote the companies covered. The thesis contributes to extant literatures on the relevance of accounting information, content of analysts’ reports, analysts’ bias and impression management. Moreover, the findings have policy implications as they speak to the concern about the relevance of accounting information and highlight the need to consider the subjective influences and the role of analysts’ incentives. Additionally, policy intervention on analysts’ bias should extend beyond recommendations and earnings’ forecast and consider the largely unregulated nature of the narrative content of the reports.