There are wide variations in the macronutrient values adopted by neonatal intensive care units and industry to fortify milk in efforts to achieve recommended intakes for preterm infants. Contributing to this is the variation in macronutrient composition of preterm milk between and within mothers and the variable quality of milk analyses used to determine the macronutrient content of milk. We conducted a systematic review of the literature using articles published in English between 1959 and 2013 that reported the concentrations of one or more macronutrients or energy content in human preterm milk, sampled over a representative 24-h period. Searched medical databases included Ovid Medline, Scopus, CINAHL and the Cochrane Library. Results are presented as mean values and ranges for each macronutrient during weeks 1–8 of lactation, and preferred mean values (g/100 ml) for colostrum (week 1) and mature milk (weeks 2–8; protein: 1·27, fat: 3·46, lactose: 6·15 and carbohydrate: 7·34), using data from studies employing the highest-quality analyses. Industry-directed fortification practices using these mean values fail to meet protein targets for infants weighing <1000 g when the fortified milk is fed <170–190 ml/kg per d, and the protein:energy ratio of the fortified milk is inadequate. This study aimed to provide additional information to industry in order to guide their future formulation of breast milk fortifiers. Quality macronutrient analyses of adequately sampled preterm breast milk would improve our understanding of the level of fortification needed to meet recommended protein and energy intakes and growth targets, as well as support standardised reporting of nutritional outcomes.