||The online reference manager tool Mendeley (http://www.mendeley.com/) is one of the most promising tools for altmetrics research (Li, Thelwall and Giustini, 2011;Wouters & Costas, 2012) and it has been already used in other previous studies, for example in Library and Information Science Journals (Bar-Ilan et al., 2012b), for Nature and Science journals (Li, Thelwall and Giustini, 2012); for PLoS ONE publications (Priem, Piwowar & Hemminger, 2012); and for a sample of all WOS disciplines (Zahedi, Costas & Wouters; 2013). The concept of Altmetrics was introduced by Priem et al. (2010) and it has been frequently referred to as an alternative way of measuring broader research impacts (other than citation) and ‘real time’ impact in social web via different tools. Most of studies investigated how altmetrics capture different type of impact compare to citations (some of them mentioned above); while in others the focus has been on how altmetrics can be used as predictor of citations. For example, in case of F1000, found that recommendations have a relatively lower predictive power in indicating high citedness as compared to journal citation scores (Waltman & Costas, 2013), moreover, correlation between F1000 labels and citation impact were not statistically signiﬁcant in most cases (Mohammadi & Thelwall, 2013); also weak correlation among users’ tag and bookmarks as an indicator of journal usage and perception and citations observed for physical journals (Haustein, & Siebenlist, 2011). In the case of Mendeley, the correlation with citations has been observed to be higher (Bar-Ilan et al., 2012a; Bar-Ilan et al., 2012b; Priem, Piwowar & Hemminger, 2012; Li, Thelwall and Giustini, 2012; Li & Thelwall, 2012; Zahedi, Costas & Wouters; 2013), however, so far the relationship of the different types of readers with the impact of the publications has not yet been explored. For this reason, in this study, we present an exploratory analysis of the patterns of reading of the different types of users in Mendeley and we study their relationship with citations. Thus, our main objective is to know if there are different patterns in terms of impact depending on the different ‘career stages’, ‘disciplines’ and ‘countries’ of the readers in Mendeley. In the case of finding different types of impact and reading patterns among Mendeley readers, this could open the door to detect different types of impact (e.g. education impact or professional impact) and even to introduce the possibility of considering the different users as potential predicting elements of citations. Methodology & preliminary results: In this research we have studied two random samples of publications from the Web of Science: the first one containing 20,000 publications published between 2005 and 2011 from all disciplines, and the second sample include 200,000 publications published between 2011 and 2012 also from all disciplines. Both gathered during March and April 2013 via the Mendeley API and using the DOI of the publications as the linking element. For the two samples we have also calculated standard bibliometric indicators (Waltman et al., 2011). For the analysis of the users we have considered the information of the top three ‘career stage users’, ‘countries’ and ‘disciplines’ of the users. We acknowledge the limitation of counting only with the top three and we discuss this in the paper. Some preliminary results show that PhD students tend to read papers with higher impact than other users and also that they read more recent papers. Further research will be done in order to explore other potential factors (e.g. the higher presence of PhD students among the users of Mendeley) that can influence this observation.