Replacing a legacy system is a complex information system (IS) change project and represents severe economic and operational risks for an organization. Due to this, legacy systems are hard to get rid of and thus they sometimes continue to live alongside their intended replacement systems, forming complex IS architectures. To understand how this happens, I conducted an in-depth case study of a factory that failed to kill its legacy system despite executing two massive IS implementations. Interviews with key stakeholders reveal that the factory has gotten caught in between old and new system architectures. Leveraging the punctuated socio-technical information system change (PSIC) model as a theoretical sensitizing device, I provide a process explanation for this outcome. Moreover, I propose extensions to the PSIC model to better capture the nuances of complex IS change. My findings help managers to make more informed decisions when engaging into complex IS change projects.