In this work a model envisaging the integrated optimization of bioreaction and downstream processing is presented. This model extends the work presented in part 1 of this pair of papers by adding ultrafiltration to process optimization. The new operational parameters include ultrafiltration time, pressure, and stirring rate. For global optimization, the model uses as constraints the final product titer and quality to be achieved after downstream processing. This extended model was validated with the same system used in part 1, i.e., PA317 cells producing a recombinant retrovirus containing the LacZ gene as a marker in stirred tanks using porous supports. Optimization of the extended model led to the conclusion that bioreaction should have two steps, batch and perfusion, similar to what was found in part 1. Ultrafiltration in a stirred cell should be performed at low pressures and stirring rates to reduce the losses of infective retroviruses. Sensitivity analysis performed on the results of the integrated optimization showed that under optimal conditions the productivity is less sensitive to the parameters related to ultrafiltration than to those associated with bioreaction. These results were interpreted as reflecting the high yield of ultrafiltration (90%). The relevance of the model extension to perform integrated optimization was also demonstrated since a restriction in the specific ultrafiltration area in downstream processing conditioned perfusion duration and perfusion rate in bioreaction. This clearly indicates that overall process optimization cannot be achieved without integrated optimization.