The aim of this thesis was to determine the effect of glucose feeding and time of induction on the productivity of a recombinant E. coli fermentation. The process chosen is currently the most common industrial recombinant fermentation process (i.e. temperature induced E. coli producing high yields of heterologous protein and using a complex feed) for which there is a scarcity of information in the literature. Two different induction times were considered: A "medium" cell density process induced at 4.5 g DCW/L which used air sparging and a "high" cell density process induced at 9.0 g DCW/L which used oxygen enriched sparging to support the higher cell densities. All experiments were performed in a 7L fermenter using a host/vector in which the bovine somatotropin (B.S.T.) gene was controlled via the cI857 temperature sensitive repressor. Results showed that there was an optimal glucose feedrate to maximise the B.S.T. titre or yield above which acetate inhibition occurred and below which the cells were starved of glucose. It was also shown that inducing at twice the cell density did not affect the specific rate of acetate production and led to the same or higher specific B.S.T. production rate. Thus the high cell density fermentation had a faster rate of B.S.T. production (since it had more cells) but also reached inhibitory concentrations of acetate sooner. As a result the final B.S.T. titre was lower in the high cell density process. As an aid to understanding and comparing the processes an unstructured mathematical model of the fermentation was developed. The model predicted the glucose, biomass, acetate and B.S.T. concentration throughout a fermentation given inoculum conditions and the profiles for glucose feeding and temperature. Uniquely the model predicts the production and inhibitory effect of acetate including the increased production of acetate due to peptone feeding. It was shown how the model could accurately predict the effects of glucose feeding but was unable to predict the effects of time of induction. However examples are shown of how the existing model could be used for process development. Significantly the model is simpler than existing mathematical models for recombinant E.coli fermentation and is therefore more likely to be useful in an industrial context.