A greenhouse experiment was carried out to study the severity of phosphorus (P) deficiency symptoms on leaves, shoot dry matter production, and shoot concentration and content (the total amount per shoot) of P in 39 bread wheat (Triticum aestivum L.) and 34 durum wheat (Triticum durum L.) genotypes grown in a severely P-deficient calcareous soil with low (20 mg P kg(-1) soil) and adequate (80 mg P kg(-1) soil) P supply for 39 days. As the seed P concentration or content can affect plant performance under P-deficient conditions, the seeds of the genotypes used in the present study were also analyzed for P concentration. Phosphorus efficiency (relative shoot growth) of genotypes, calculated by the ratio of shoot dry matter production under low P to that under adequate P supply, significantly differed among the genotypes, and varied between 46.7% and 78.6%. Phosphorus efficiency ranged from 51% to 71% with an average of 61% for bread and from 47% to 79% with an average of 66% for durum wheat genotypes. There was no correlation between P efficiency ratio and P concentration of plants (R-2 = 0.0001), but P efficiency of all bread and durum wheat genotypes showed a very significant correlation with the P content (the total amount of P per shoot) (R-2 = 0.333***). The relationship between the P efficiency and total amount of P per shoot was much more significant in bread (R-2 = 0.341***) than in durum wheat (R-2 = 0.135*). Like shoot P concentrations, also severity of visible leaf symptoms of P deficiency on older leaves, including leaf chlorosis and necrosis, did not correlate with P efficiency. In most cases, genotypes showing higher P efficiency had higher absolute shoot dry weight under P deficient conditions. Under P deficient conditions, the absolute shoot dry weight very significantly correlated with shoot P content (R-2 = 0.665***), but the correlation between the absolute shoot dry weight and shoot P concentration tended to be negative. There was also variation in native seed P reserve of the genotypes, but this variation had no influence on the P efficiency. The results indicate that the total amount of P per shoot and shoot dry matter production at low P supply are most reliable parameters in ranking genotypes for P efficiency at early growth stage. In wheat germplasm tested in the present study, several wheat genotypes are available showing both very high P efficiency and very high shoot content and concentration of P suggesting that P acquisition ability should be most important mechanism for high P efficiency in such genotypes. On the other hand, there are also genotypes in the germplasm having more or less same P concentration or P content in shoot but differing substantially in P efficiency, indicating importance of P utilization at cellular level in P efficiency. All these results suggest that P efficiency mechanisms can be different from one genotype to other within a given plant species.