The amount of recombinant product obtained from mammalian cells grown in a bioreactor is in part limited by achievable cell densities and the ability of cells to remain viable over extended periods of time. In an attempt to generate cell lines capable of better bioreactor performance, we subjected the DG44 Chinese Hamster Ovary (CHO) host cell line and a recombinant production cell line to an iterative process whereby cells capable of surviving the harsh conditions in the bioreactor were selected. This selective process was termed "bioreactor evolution". Following the selective process, the "evolved" host cells attained a 2-fold increase in peak cell density and a 72% increase in integral cell area. Transient transfection experiments demonstrate that the evolved cells have the same transfection efficiency and the same secretory potential as the initial cells. The "evolved" host was also found to contain a large subpopulation of cells that did not require insulin for growth. From this, a new population of growth-factor-independent cells was obtained. These improvements in host properties should prove beneficial in the expression of recombinant proteins in fed-batch processes. The selective process was also applied to a recombinant production cell line. The evolved cells from this selection exhibited a 38% increase in peak cell density, a 30% increase in integral cell area, and a 36% increase in product titer. These increases were obtained without any appreciable impact on product quality, demonstrating the usefulness of this simple approach to improve the performance of recombinant cell lines.