The cyclic AMP (cAMP) response element-binding protein (CREB) has been demonstrated to be a key mediator of cellular promoter response to cAMP. The binding site for this protein in many cellular cAMP inducible promoters (CRE) contains the palindrome sequence TGACGTCA, which contains two half-sites for CREB binding. A related promoter element, with the core sequence TGACG, has significant homology to an AP1-binding site and contains only one half-site for CREB binding. A group of factors known as activating transcription factors (ATF) have been found to bind to the latter and related sequences found upstream of early adenovirus promoters induced by E1A, and these factors are highly homologous to the CREB protein. We wished to characterize CREB, c-jun, and c-fos binding to these sites in the somatostatin gene (CRE) and in the adenovirus early region 3 promoter (E3/ATF). Oligonucleotides complementary to each of these sites were used in gel retardation assays with in vitro-translated CREB protein. These studies indicated that CREB bound primarily as a dimer to both a single and two half-sites, though there was increased affinity to the double compared with the single half-site. The c-jun and c-fos proteins also bound to both the somatostatin CRE- and E3/ATF-binding sites, but CREB did not bind to AP1 recognition sites nor was it capable of forming heterodimers with either c-jun or c-fos. Truncations of the CREB protein, which eliminated regions of the protein containing consensus sites for phosphorylation by protein kinase A, protein kinase C, and casein kinase II, bound to both the CRE and ATF sites, indicating that these consensus sites were not essential for DNA binding or dimer formation. Transfection of CREB and protein kinase A expression constructs into F9 cells with promoters containing either a single or two half-sites for CREB binding indicated that CREB was capable of similar levels of activation of these constructs. However, the fold activation by CREB was higher for constructs containing a single half-site compared with those containing two half-sites. These results demonstrate that multiple mechanisms may regulate CREB binding, including variations in the sequences in the promoter-binding site and the presence of related DNA-binding proteins.