The divergent homeodomain gene pdx1 plays an essential role in vertebrate pancreas development. Loss of pdx1 function in rodents and humans perturbs growth and differentiation of the pancreas primordium and results in pancreatic agenesis (Jonsson et al., 1994; Offield et al., 1996; Stoffers et al., 1997). pdx1 is not required during the early stages of pancreas development as specification of the prepancreatic gut endoderm is normal in pdx1 mutants (Offield et al., 1996; Ahlgren et al., 1996). The pancreatic bud develops a primitive ductular architecture in the absence of pdx1 gene function (Offield et al., 1996; Ahlgren et al., 1996), although subsequent endocrine and exocrine development is severely impaired. As a result, pdx1 mutants lack functional pancreatic tissue (Jonsson et al., 1994; Offield et al., 1996), and it has been suggested that pdx1 regulates pancreatic stem cell function. Consistent with this hypothesis, pdx1 expression is markedly increased in proliferating ductular cells during pancreatic regeneration (Kritzik et al., 1999; Sharma et al., 1999). Although essential for early pancreas development, pdx1 is not sufficient for pancreatic organogenesis (Grapin-Botton et al., 2001; Heller et al., 1998). When expressed ectopically, pdx1 rarely induces b-cell gene expression (Ferber et al., 2000). There are, however, considerable data showing that pdx1 regulates endogenous b-cell function during development, and in adults, through its transcriptional activation of genes such as insulin and glut-2 (Peshavaria and Stein, 1997; Peshavaria et al., 2000). Consistent with this role, reduced pdx1 activity is associated with diabetes in animal models and humans (Stoffers et al., 1997; Ahlgren et al., 1998; Habener and Stoffers, 1998; Peshavaria et al., 2000). The role of pdx1 in zebrafish pancreas development was analyzed in pdx1 morphants. When injected at the one-cell stage, a pdx1 morpholino perturbed pancreas development in a dose-dependent manner (Table 1A, B). When analyzed at 48 h postfertilization (hpf), pdx1 morphants had minimal immunoreactive insulin, glucagon, or somatostatin, whereas buffer-injected controls had a solitary prominent islet comprising all three hormoneexpressing cells (Fig. 1A–H). pdx1 morphants examined at this stage also had minimal exocrine tissue, based on the expression levels in whole mounts of gata-6, an established endodermal marker (not shown). At 72 hpf, the pancreatic islet of pdx1 morphants was smaller than buffer-injected controls (Fig. 1J, N) and there was minimal exocrine tissue, based on the number of cells expressing gata-6 (Fig. 1K, O) or immunoreactive for carboxypeptidase-A (Fig. 1J, N), a marker of differentiated acinar cells. Tissue sectioning confirmed these findings (Fig. 1L, P) and also showed that the main pancreatic duct formed in pdx1 morphants (not shown). Specificity of these results was confirmed via rescue of the pdx1 morphant pancreatic phenotype by microinjected zebrafish pdx1 mRNA, and the presence of a normal pancreas in 48 hpf hhex and cyclooxygenase2 morphants (not shown). The reduction in pancreatic tissue seen in pdx1 morphants is consistent with the function of mammalian pdx1 (Jonsson et al., 1994; Offield et al., 1996). However, morpholino-mediated knockdown of pdx1 activity was reversible as both the endocrine and exocrine pancreas of 5-day postfertilization (dpf) pdx1 morphants appeared normal (Fig. 1Q–X). To determine whether the mechanism underlying pancreatic recovery resembled mammalian pancreatic regeneration (Kritzik et al., 1999; Sharma et al., 1999), we analyzed cell proliferation in 48, 60, and 72 hpf pdx1 morphants using 5-bromo-29-deoxyuridine immunohistochemistry. These data revealed that proliferating cells in the pancreas of pdx1 morphants were present within tissue surrounding the islet in the region of the developing exocrine pancreas but never within the islet (Fig. 1Z), as in identically staged wild-type zebrafish (Fig. 1Y) and the regenerating pancreas of mammals (Sharma et al., 1999).