In Hordeum vulgare, phosphorus retranslocation was studied after it had been supplied to the roots for three days (experiment 1), and after foliar application (experiments 3–8). Phosphorus uptake by leaves of different ages was also measured 16 and 60 minutes after 32P addition to the medium (experiment 2). In experiment 1, treatments at 0.6 and 31 p.p.m. of phosphorus were applied when the first leaf had completed its rapid growth. The plants were then grown for three days in media labelled with 32P, and for a subsequent 10 days in non-labelled solutions. Retranslocation was measured by changes in total phosphorus and in 32P. Both root feeding, and foliar application of 32P, demonstrated three phases during leaf development: import (recently initiated leaf), export (mature leaf) and an intermediate phase with both export and import (leaf half developed). There was large transport of foliar applied 32P, from mature leaves to roots, and some of this 32P was re-exported to the shoots, including the mature leaves. Root feeding of 32P over short periods strongly suggested that phosphorus uptake by the shoots occurred via the xylem, even at low phosphorus. In experiment 1, there were distinct treatment differences in relative growth rates, growth of young organs and roots, and in phosphorus concentrations of all but the very young leaves. Mature leaves showed a large net phosphorus export at low phosphorus, but a large net import at high phosphorus. This was not due to treatment differences in export, because total export from the mature leaves was even somewhat smaller at low than at high phosphorus. The treatment differences, with net export at low but net import at high phosphorus, were thus due to the higher import in the mature leaves at high phosphorus. Total export remained at a high level throughout the experiment at high phosphorus, while it declined with time at low phosphorus. For phosphorus absorbed during early growth, both the export from the mature leaves, and the intake by the developing leaves, was independent of phosphorus treatment; i.e. for each individual organ the quantities of phosphorus involved were the same in the two phosphorus treatments. Thus, the higher phosphorus contents of developing organs at high phosphorus were obtained from phosphorus supplied to the roots during later growth, and not from phosphorus supplied during early growth of the whole plant. The data are consistent with the notion that phosphorus export is controlled in the source. It is suggested that at high phosphorus this control is due to a saturation of the sites transporting phosphorus into the phloem. At low phosphorus, on the other hand, release from individual leaf cells might have been the dominating factor.