Implantation of encapsulated nonautologous cells that have been genetically modified to secrete proteins with tumor suppressor properties represents an alternative nonviral strategy to cancer gene therapy. We report an approach to raise the yield of recombinant proteins from encapsulated cells substantially. We hypothesized that by optimizing the encapsulation procedure, the production efficacy from the encapsulated cells could be increased. HEK 293 EBNA cells were genetically engineered to produce angiostatin. Encapsulation was performed by varying bead size, cellular density, homogeneity, and ion composition of the gel. The morphology and viability of the cells and the release of angiostatin were studied. Computer software was developed for three-dimensional imaging and quantification of cell viability. Angiostatin production was assessed at 3, 6, and 11 weeks using enzyme-linked immunosorbent assay (ELISA). Inhomogeneous gels facilitated cell growth and viability. The most efficient inhomogeneous microcapsules were generated by reducing the size and cellular density of the beads. The viability and the production of angiostatin were 3 to 5 times higher than in the homogeneous capsules. Significant amounts of viable cells were present in both homogeneous and inhomogeneous beads after 6 months of culture. The stability of the alginate matrix was greatly enhanced by gelling in the presence of barium. In conclusion, the viability and production efficacy of recombinant angiostatin from alginate-encapsulated cells can be increased considerably by optimizing the encapsulation procedure. The development of such optimized microcapsules brings cell-encapsulation therapy further towards clinical use in cancer therapy.