IL15, a member of the common gamma chain receptor (gamma c) cytokine family, is gaining attention in recent years as one of the most promising anti-tumor agents. IL15 regulates T cell activation and proliferation, promotes the survival of CD8(+) CD44(hi) memory T cells and is also essential for NK cell expansion and development. Despite the attraction of developing IL15 as an anti-cancer agent, production of recombinant IL15 has proven to be difficult due to the stringent control of IL15 expression at the transcriptional, translational and the post-translational levels. Furthermore, the bioactivity of IL15 fused to an extra functional domain that is isolated from mammalian cells is generally inferior to recombinant IL15 produced by E. coli. In this study, we report that Lysine 86 in IL15 is responsible for the instability in mammalian cells when its C-terminus is fused to the albumin binding scFv (IL15-A10m3). We demonstrate that K86A or K86R mutants increased the expression of the fusion protein from HEK293 cells. When the wild type IL15 is used for the fusion, no recombinant IL15 fusion was detected in the culture media. Additionally, we determined that the residue 112 in IL15 is critical for the bioactivity of IL15A10m3. Examination of single and double mutants provides a better understanding of how IL15 engages with its receptor complex to achieve full signaling capacity. The results of our experiments were successfully applied to scale up production to levels up to 50mg/L and > l0mg/L of > 95% pure monomelic recombinant fusion proteins after a 2-step purification from culture media. More importantly, the recombinant fusion protein produced is fully active in stimulating T cell proliferation, when compared to the recombinant wild type IL15.