Computer-supported collaborative learning (CSCL) environments, particularly environments where students share external representations, are discussed as an interesting area for the application of cognitive load theory (CLT). CSCL environments share a number of characteristics that will induce considerable cognitive load. This article reflects on the characteristics of CSCL environments and external representations and the research opportunities they offer to CLT. We argue that CLT can contribute to our understanding of CSCL by pinpointing situations in which high levels of cognitive load are generated. CLT also raises a number of particular issues: from a CLT point of view studying of worked out external representations may lead to better learning outcomes than (co)-construction which is the standard mode of operation in CSCL. The ontology and specificity of the representational scheme used are crucial factors, as they define what can be expressed and how much room for interpretation is left in the representation. CLT may help identify mismatches between representational schemes and execution of (sub)tasks. CLT has lead to a number of recommendations on the integration of diagrammatic and textual information. We relate this to discussion and chat facilities in CSCL and describe how these can be more firmly anchored in the external representation.