Proteins fulfil a wide variety of essential functions in the cell. Recombinant protein production in the Gram-negative bacterium Escherichia coli (E. coli) facilitates structural and functional studies of proteins and it has been instrumental in biotechnology for the manufacturing of e.g. many protein-based drugs. However, to obtain sufficient amounts of active recombinant protein is not always a trivial task. The production of proteins that reside in membranes is limited by their complex biogenesis and their hydrophobic nature. Consequently, despite the importance of membrane proteins in health and disease (approx. 70% of today’s drugs act on membrane proteins), structures of only a very small fraction of the existing membrane proteins have yet been solved. Many soluble proteins are also difficult to produce, e.g. those containing disulfide bonds (e.g. antibody fragments and most hormones). Disulfide bond-containing proteins have to be produced in the periplasm of E. coli to be able to fold properly. To reach the periplasm, these proteins have to be ‘secreted’ across the inner membrane, which makes that also their biogenesis is complex. The aim of my PhD studies has been to improve E. coli-based production of recombinant membrane and secretory proteins. I have found that (i) the previously developed Lemo21(DE3) protein production strain can be used to set the expression intensity of a gene encoding a membrane protein such that the protein is optimally produced in the cytoplasmic membrane without causing any notable stress. Also, (ii) membrane protein production using the Lemo21(DE3) strain can be improved and simplified using carefully optimized culturing and induction conditions, demonstrated by the development of the ‘MemStar recipe’. Furthermore, I found that (iii) when using the standard BL21(DE3)/pT7 expression system for the production of membrane and secretory proteins, omitting the inducer IPTG leads to drastically improved yields as compared to when IPTG is added, owing to a lower initial target protein production rate. In the fourth study, I found that (iv) when using the PrhaBAD promoter for expression of the target gene, protein accumulation rates appear to be mostly unaffected by the inducer concentration. Using a strain-engineering approach, PrhaBAD-based protein production rates could be made constant and rhamnose concentration dependent. This dramatically improved production yields of both membrane and secretory proteins, using only very low amounts of inducer. Taken together, in accordance with previous studies, lowering production rates is an efficient strategy to increase production yields of both membrane- and secretory proteins. This is mostly due to alleviating saturation of the machinery involved in the biogenesis of these proteins. Finally, I also conducted a study (v) where, in both E. coli and Salmonella, I orchestrated the production of two membrane proteins (one that mediates the production of antigens on the surface of Gram-negative bacteria and another that makes defined pore-structures in the Gram-negative bacterial cell envelope) for the development of a safe (non-living) vaccine platform.