Directors execute trades in their companies’ shares for a variety of reasons. Their actions may have information value for other investors seeking to mimic these strategies. This paper tests whether directors’ discretionary trading has a predictive value, and also provides a better understanding of the information flow in equity markets. This paper aims to better understand the informational content of trades executed by directors, as these trades are meaningful to other market participants. Directors trade for a variety of reasons, and through different channels. For example, they may make direct on-market trades, or may trade indirectly by way of participation in dividend reinvestment plans. The distinction between these trades is the discretion exercisable over the key trading parameters of the timing and size of the trade. By distinguishing between different types of trades, this paper tests whether trade discretion has a predictive value. Additional light is shed on the characteristics of the most informative trades, as well as the interaction between directors’ trading and that of other market participants. The study has two key findings. First, trades that are made with complete discretion are significantly more informative than other trades. Second, trades are more informative as the level of discretion increases. The underlying thinking is that the greater the ability to incorporate information into a trade, the more informative it tends to be. Unsurprisingly, firms in the sectors most susceptible to information asymmetry on average feature more informative trades. Direct and indirect interests are compared, and purchases are found to be more informative for indirect, as opposed to direct, transactions. The opposite appears true for sales. This suggests a contrast in directors’ behaviour when buying as opposed to selling. More specifically, evidence indicates that purchases are driven largely by fundamental factors, while sales are more event-driven. This behaviour is demonstrated by earnings announcements. Directors appear to decrease discretionary purchases following positive announcements, whilst discretionary sales are predictive of negative announcements. Analyst stock coverage is also found to influence the buying behaviour of directors. Both a higher consensus rating and an upgraded rating result in fewer discretionary purchases, with no evident effect on sales. Moreover, the breadth of analyst coverage is a proxy for information asymmetry. Directors’ trades become less informative as the number of analysts covering the firm increases. Finally, other market participants are found to be able to distinguish between informative and non-informative director trades. Evidence of institutional investors mimicking directors’ trades, and bid-ask spreads widening on the day of these transactions, is significantly stronger than for other trades. These results affirm the importance of identifying particular director trades that are most likely to contain information value. The implications of the simple method used in this paper provide a better understanding of the information flow in equity markets, and indicate further research opportunities to identify meaningful signals from the discretionary trades of directors. Evidence of a behavioural finance framework for directors’ trades is also provided, implying that directors are not immune to these behavioural biases.