This study was performed to test if alternative carbon sources besides recently photosynthetically fixed CO2 are used for isoprene formation in the leaves of young poplar (Populus x canescens) trees. In a 13CO2 atmosphere under steady state conditions, only about 75% of isoprene became 13C labeled within minutes. A considerable part of the unlabeled carbon may be derived from xylem transported carbohydrates, as may be shown by feeding leaves with [U-13C]Glc. As a consequence of this treatment approximately 8% to 10% of the carbon emitted as isoprene was 13C labeled. In order to identify further carbon sources, poplar leaves were depleted of leaf internal carbon pools and the carbon pools were refilled with 13C labeled carbon by exposure to 13CO2. Results from this treatment showed that about 30% of isoprene carbon became 13C labeled, clearly suggesting that, in addition to xylem transported carbon and CO2, leaf internal carbon pools, e.g. starch, are used for isoprene formation. This use was even increased when net assimilation was reduced, for example by abscisic acid application. The data provide clear evidence of a dynamic exchange of carbon between different cellular precursors for isoprene biosynthesis, and an increasing importance of these alternative carbon pools under conditions of limited photosynthesis. Feeding [1,2-13C]Glc and [3-13C]Glc to leaves via the xylem suggested that alternative carbon sources are probably derived from cytosolic pyruvate/phosphoenolpyruvate equivalents and incorporated into isoprene according to the predicted cleavage of the 3-C position of pyruvate during the initial step of the plastidic deoxyxylulose-5-phosphate pathway.