Endostatin is a potent angiogenic inhibitor that has been approved for the treatment of cancer. However, endostatin is unstable in vitro and difficult to produce in large quantities. Endostatin gene therapy is an alternative to overcome these difficulties by expressing sustained bioactive endostatin in vivo. We previously developed a recombinant replication-defective adenovirus, E10A, which carries the human endostatin gene. Phase I trial of E10A given as a weekly intratumoral injection in adult patients with solid tumors has been finished. The clinical application of endostatin gene therapy was limited by the high cost of large-scale production. In the current study, we found that there was a high level (100 mg/L) of endostatin in the fermentation supernatant of 293 cells transfected with Ad/rhEndo. A protocol was developed to purify recombinant endostatin in the fermentation supernatant to a yield of 24 mg/L and 98% purity by the use of SP Sepharose FF cation exchange chromatography, Sepharose-heparin Hi Trap affinity chromatography and gel filtration chromatography. The anti-proliferative activity of 293 cell-expressed endostatin (ArhEndo) is comparable to that of yeast-expressed endostatin (YrhEndo). Cell migration assay indicated that ArhEndo is more effective than YrhEndo. Moreover, ArhEndo is of higher stability than YrhEndo. These results suggested that purification of recombinant endostatin from fermentation supernatant provided an economic and available strategy for Ad/rhEndo production. (C) 2007 Elsevier Inc. All rights reserved.