The literature documents a pattern of positive returns around earnings announcements (EAs), commonly referred to as the earnings announcement premium (EAP). Studies either cite an information explanation, attributing this phenomena to (i) increased risk or market frictions because of the information revealed in the EA (Ball and Kothari 1991; Cohen et al. 2007; Patton and Verardo 2012; Levi and Zhang 2015; Savor and Wilson 2016), 1 or (ii) an attention explanation, because of greater investor attention around the announcement (Frazzini and Lamont 2007; Hirshleifer et al. 2008; Aboody et al. 2010; Lawrence et al. 2017). Recent research mainly supports the information view and questions the robustness of evidence for the attention explanation. These studies suggest that results supporting the attention explanation conflate the effects of information and attention, use empirical proxies that do not explain positive returns before the EA (when much of the EAP is generated), and are sensitive to variable construction choices (Cohen et al. 2007; Barber et al. 2013; Levi and Zhang 2015). Additionally, these two explanations have been mostly studied independently, with papers supporting one explanation largely overlooking the other. Research thus fails to describe how they might jointly explain the EAP. The objectives of this paper are to provide new evidence of the attention explanation and reconcile the two explanations. I proceed in three phases. First, I document an attention effect, similar to the EAP, for earnings notifications, which points to an attention effect for EAs, given their similar attention-grabbing properties. Second, I show that earnings notifications attenuate the attention-grabbing effect of EAs, allowing for empirical identification of the effect of attention on the EAP. Third, I provide further evidence of the attention effect and compare it to the information-based explanations, showing how both explain the EAP. In the first phase, I use a setting of earnings notifications, which, like EAs, are firm-initiated press releases distributed to investors via paid newswires and are therefore likely to grab attention. However, unlike EAs, notifications provide no additional information about firm performance. Thus I argue they provide a research setting that disentangles the effect of attention from that of new information. 2 I find abnormal returns of between 8 and 11 basis points, 41% more internet searches of the SEC’s repository of firm filings (EDGAR searches), and 5% higher trading volume on notification days. As additional evidence of an attention effect, I show that notification-day returns are higher for lower visibility firms (where additional attention should have a larger effect) and firms for which short selling is more constrained (which limits the trading choices to buying or abstaining from trade, increasing the effect of attention on returns). In the second phase, I show that earnings notifications attenuate attention to the EA. Given the similar attention-grabbing properties of notifications and EAs and the fact that the earnings notification occurs soon before the EA, the behavioral literature on habituation predicts a diminished response to the EA (Thompson and Spencer 1966; Groves and Thompson 1970; Rankin et al. 2009). Thus, to the extent that the EAP is caused by attention, the effect should diminish in the presence of an earnings notification. Consistent with this prediction, I find that, when firms provide notifications, returns are 0.24% higher over the five trading days before the EA (when notifications often occur) and 0.22% lower around the EA. EDGAR searches around the EA are also lower in the presence of earnings notifications.