We investigated the mechanism of amylose synthesis in Arabidopsis leaves using C-14-labeling techniques. First, we tested the hypothesis that short malto-oligosaccharides (MOS) may act as primers for granule-bound starch synthase I. We found increased amylose synthesis in isolated starch granules supplied with ADP[C-14]glucose (ADP[C-14]Glc) and MOS compared with granules supplied with ADP [14C]Glc but no MOS. Furthermore, using a MOS-accumulating mutant (dpe1), we found that more amylose was synthesized than in the wild type, correlating with the amount of MOS in vivo. When wild-type and mutant plants were tested in conditions where both lines had similar MOS contents, no difference in amylose synthesis was observed. We also tested the hypothesis that branches of amylopectin might serve as the primers for granule-bound starch synthase I. In this model, elongated branches of amylopectin are subsequently cleaved to form amylose, We conducted pulse-chase experiments, supplying a pulse of ADP[C-14]Glc to isolated starch granules or (CO2)-C-14 to intact plants, followed by a chase period in unlabeled substrate. We detected no transfer of label from the amylopectin fraction to the amylose fraction of starch either in isolated starch granules or in intact leaves, despite varying the time course of the experiments and using a mutant line (sex4) in which high-amylose starch is synthesized. We therefore find no evidence for amylopectin-primed amylose synthesis in Arabidopsis. We propose that MOS are the primers for amylose synthesis in Arabidopsis leaves.