We have developed a serum-free cell culture process utilizing a recombinant baculovirus (AcNPV) expression vector to infect Trichoplusia ni insect cells for the production of the human lysosomal enzyme, glucocerebrosidase. The enzyme, which is harvested as a secreted protein in this process, can serve as a replacement therapy for the genetic deficiency Gaucher disease. In the course of pilot scale-up of a batch glucocerebrosidase process from 25-mL working volume shaker flask units to 25-L working volume stirred bioreactor units, a semiempirical model was developed for the rational determination of scaleable process parameters, including host cell density at infection, multiplicity of infection (MOI), and harvest time. A key assumption of the model is that maximum protein production is limited by the serum-free medium’s nutritional capacity, which can, in turn, be determined from the growth of uninfected cells. For the host cell/medium combination used in this study, the nutritional limit was determined to be 1.3 x 10(7) to 1.7 x 10(7) viable-cell-days/mL. Based on this, the model predicts that optimal protein expression is consistent with a 4-day batch process where the host cell density at the time of infection is 1.5 x 10(6) to 2.0 x 10(6) cells/mL and the MOI is 0.09-0.3. These parameters were empirically confirmed to give the highest achievable batch product yield, first in shaker flasks and then at larger scales. The tow MOI allows at least one population doubling to take place post viral addition, so that the effective infected cell density producing product generally exceeds 4 x 10(6) cells/mL. It was also interesting to note that this process consistently achieved the same level of maximum protein production at the 25-L bioreactor scale in 4 days compared to 5 days at the shaker flask scale. This may be attributable to better control of the culture environment in the bioreactor. Unlike some other lepidopteran insect cells, such as Sf-9, T. nj cells were found to produce significant levels of the inhibitory metabolites ammonia and lactate. Our results suggest that reduction and/or removal of inhibitory metabolites might be beneficial for infection of high-density cultures of these cells and might also facilitate application of more sophisticated culture strategies, including fed-batch.