In multimedia content sharing social networks, tags assigned to content play an important role in search and retrieval. In other words, by annotating multimedia content, users can associate a word or a phrase (tag) with that resource such that it can be searched for efficiently. Implicit tagging refers to assigning tags by observing subjects behavior during consumption of multimedia content. This is an alternative to traditional explicit tagging which requires an explicit action by subjects. In this paper we propose a brain-computer interface (BCI) system based on P300 evoked potential, for implicit emotional tagging of multimedia content. We show that our system can successfully perform implicit emotional tagging and naïve subjects who have not participated in training of the system can also use it efficiently. Moreover, we introduce a subjective metric called "emotional taggability" to analyze the recognition performance of the system, given the degree of ambiguity that exists in terms of emotional values associated with a multimedia content.