The Internet has become an increasingly important source of science information for laypersons. Users have access to a virtually unlimited array of documents by diverse sources, which offers enhanced opportunities of relevant information acquisition but also leads to new challenges: First, not all documents adequately represent the scientific state of the art. Second, not all readers have the skills to understand complex scientific information. Particularly in the domain of current scientific debates, for instance on the effects of emerging technologies, the state of research is characterized by complex patterns of conflicting evidence, which may be difficult to understand for laypersons. Against this background, the talk deals with the question of how laypersons select and process science articles (varying in message complexity and source descriptions) in new media environments such as science blogs. Experimental results indicate that laypersons for whom a topic is relevant are relatively open-minded toward two-sided (more complex) messages on conflicting evidence – however, people with naive beliefs on the nature of science and lower need for cognition tend to process unbalanced information in a biased way. The presentation will present empirical findings and will discuss theoretical and practical implications regarding laypersons' use of online science information.