Genetic improvement in phosphorus (P) use efficiency (i.e. the ratio of biomass or yield at nil P to that at a given rate of application) is an important goal to improve P recovery and P efficiency of farming systems. Experiments were conducted at three sites in South Australia between 2009 and 2011 to characterise genetic variation in yield with no applied P and in the response to P fertiliser among a diverse range of barley (Hordeum vulgare L.) genotypes. In each experiment, 39–54 genotypes were grown at 0 or 30 kg P/ha. Responses to P were measured near the beginning of stem elongation by using normalised difference vegetation index (NDVI) and by harvesting the grain. Rhizosheath size was also measured on seedlings. Consistent differences in growth and yield at 0 kg P/ha were measured among the genotypes. By contrast, there were large environmental effects on responses to P, but some genotypes showed consistent responses. Measurements of growth, yield and P uptake on a subset of genotypes showed that most of the variation in biomass and yield could be attributed to variation in P-uptake efficiency (net total P uptake per unit available P) rather than to P-utilisation efficiency (biomass or yield per unit total P uptake). The size of the rhizosheath made a small contribution to variation in NDVI but not grain yield, suggesting that rhizosheath size may be of some benefit to early growth but that this does not persist through to yield. Genetic correlations between NDVI and yield were often weak but were generally positive at 0 kg P/ha. Correlations between responses in NDVI and responses in grain yield were low and often negative. The study identified several barley genotypes that showed consistent differences in yield at low P and responses to P; however, selection for P efficiency based solely on responses in vegetative growth may not be appropriate. Variation in P uptake appeared to be more important than P-utilisation efficiency for P efficiency in barley.