The aim of this review is to assess to what extent the biological processes of milk protein synthesis and secretion might be manipulated to the benefit of the milk producer, processor and consumer. There is considerable impetus for such an analysis. A completely new ‘agroceutical’ industry is in prospect, utilising novel biotechnology to produce high-value, non-mammary proteins in the milk of transgenic dairy animals (Carver et al., 1993). Nearer to home, the milk processing industry and the consumer may wish to see more modest modification of endogenous milk proteins, making use of naturally occurring genetic variants or, again, using biotechnological approaches (Dalgleish, 1992). Exciting as such possibilites are, I intend to set them to one side and concentrate on the issue that most immediately concerns the dairy farmer: how can he increase his milk protein concentration and, thereby, his profitability. Retrospective analysis reveals some small scope for improvement; between 1985 and 1990, compositional quality data (the basis for milk payment) revealed an increase in protein content from 32.6 to 32.8 g/1, an improvement of 0.6%. Over the same period, milk fat increased by 3.1% and recorded milk yields by 6.4% (all data from UK Dairy Facts and Figures, Federation of United Kingdom Milk Marketing Boards), suggesting that protein content is least amenable to manipulation. Gross production figures of this sort can be misleading, but more controlled experimentation supports the contention. By reducing the forage to concentrate ratio, Rook et al. (1992) increased total protein yield by an impressive 200 g/d. However, only one-quarter of this effect was due to improved protein content, the remainder coming from increased milk yield. Endocrine manipulation of milk production gives a similarly discouraging picture. Growth hormone treatment undoubtedly increases protein yield, but it does so by increasing the volume of milk produced with no detectable effect on protein content (Bauman, 1992). Milk production is also responsive to milking frequency but, once again, the effect is on yield of milk rather than on the protein content of that milk (Hillerton et al., 1990). So, where is the silver lining? Does biotechnology have the answer; can the presence, in mammary cells, of extra copies of correctly expressed milk protein or foreign protein genes markedly increase protein content? In Edinburgh, transgenic sheep are producing milk containing as much as 30 g/1 of human alpha- 1-antitrypsin (Carver et al.. 1993), representing some 50% or the total protein content (Wright et al., 1991). There is good anecdotal reason for believing that total protein content is elevated (one sheep reached 70 g/1), but the definitive statistical analysis is, so far, lacking. We have examined mammary function in transgenic mice expressing the sheep beta-lactoglobulin gene and compared them with non-transgenic controls (Wilde et al.,1992). Expression was good: beta-lactoglobulin accounted for 30% of all the protein present. But, at 95 and 106 mg/ml. respectively, transgenic mouse milk contained no more total protein than that of control mice. The encouraging feature of this work was that mammary tissue from transgenic mice cultured in vitro synthesised 41 % more total protein than did tissue from control mice, a significant difference (P ˂ 0.05).It would appear that the tissue itself had the intrinsic capability for increased protein synthesis, but a block or some sort was imposed in vivo. The next step is to discover the nature of that block, and to do so we must understand the complex series of events leading ultimately to protein secretion into the alveolar lumen.