Requirements engineering (RE) practices are critical to success during the development of business software. As managers assess RE practices, they apply specific perspectives that determine problems identified and recommendations for improvement. Two perspectives have recently dominated managerial thinking within the software industry, one rooted in software process improvement and the other rooted in agile software development. Underpinning these perspectives are two theories about what constitutes good software practice. In this paper, we explicate these theories in relation to RE and show how they differ in basic assumptions about the nature of requirements, requirements capture, requirements usage, change management, and approach to improvement. The repeat-ability theory holds that good requirements practices are plan-driven and follow generic best practices to arrive at an agreed-upon baseline of software requirements. Response-ability holds that good requirements practices are adaptive and involve close interaction between customers and developers to arrive at satisfactory software solutions. We use case study data from a software firm, TelSoft, to show how the theories lead to different interpretations about why current practices are problematic and how problems are resolved. Relating to the improvement strategy adopted at TelSoft, we demonstrate the superiority, for managers, of negotiating response-ability and repeat-ability concerns when improving RE practices. The paper concludes with a discussion of implications for research and practice.