This paper argues that key findings from computer-assisted language learning (CALL) research need to be replicated to permit the construction of a valid and reliable evidence-base which can inform the design of future CALL software and activities, together with language teachers' decisions about their adoption. Through the critical examination of two key studies, the limitations of much current CALL research with respect to the construction of such an evidence-base are highlighted and approaches to replication which may overcome these limitations are discussed. Like much CALL research, both papers are experimental studies. Key issues raised by the two studies are the need to: (1) conduct replications focused on attributes and affordances which transcend multiple technologies which might be employed in CALL, (2) collect process data to increase the explanatory power of studies, and (3) link CALL research to relevant second language acquisition (SLA) theory and research to increase the explanatory power of studies and integrate the findings from individual studies into the CALL and SLA evidence-base more broadly. Emphasis is also placed on the need to conduct replications with different populations in different settings in order to demonstrate the validity and reliability of current findings.