Abstract: This paper tests whether sell-side analysts are prone to behavioural errors when making stock recommendations as well as the impact of investment banking relationships on their judgments. In particular, we analyse their report narratives for evidence of cognitive bias. We find first that new buy recommendations on average have no investment value whereas new sell recommendations do, and take time to be assimilated by the market. We also show that new buy recommendations are distinguished from new sells both by the level of analyst optimism and representativeness bias as well as with increased conflicts of interest. Successful new buy recommendations are characterised by lower prior returns, value stock status, smaller firms and weaker investment banking relationships. On the other hand, successful new sells do not differ from their unsuccessful counterparts in terms of these measures. As such, we provide evidence that analysts are prone both to behavioural bias as well as potential conflicts of interest in their new buy stock recommendation decisions. We also show that these two explanations of analyst behaviour are to a great extent independent of each other. Consequently, the recent attempts by regulators to address potential conflicts of interest in analyst behaviour may have only limited impact.