We have examined the regulation of three early developmentally regulated genes in Dictyostelium. Two of these genes (D2 and M3) are induced by pulses of cAMP and the other (K5) is repressed. Expression of these genes has been examined in a number of developmental mutants that are specifically blocked in various aspects of the signal transduction/cAMP relay system involved in aggregation and control of early development. The mutant strains include Synag mutants, which are blocked in receptor-mediated activation of adenylate cyclase and do not relay cAMP pulses; FrigidA mutants, which are blocked in receptor-mediated activation of both adenylate cyclase and the putative phosphoinositol bisphosphate (PIP2) turnover pathway and appear to be mutations in the gene encoding one of the G alpha protein subunits; and a StreamerF allele, which lacks cGMP-specific cGMP phosphodiesterase. From the analysis of the developmental expression of these genes under a variety of conditions in these mutant strains, we have drawn a number of conclusions concerning the modes of regulation of these genes. Full induction of D2 and M3 genes requires cAMP interaction with the cell surface receptor and an "oscillation" of the receptor between active and adapted forms. Induction of these genes does not require activation of the signal transduction pathway that leads to adenylate cyclase activation and cAMP relay, but does require activation of other receptor-mediated intracellular signal transduction pathways, possibly that involving PIP2 turnover. Likewise, repression of the K5 gene requires pulses of cAMP. Expression of this gene is insensitive to cAMP pulses in FrigidA mutants, suggesting that a signal transduction pathway is necessary for its repression. Results using the StreamerF mutant suggest that the rise in cGMP in response to cAMP/receptor interactions may not be directly related to control of the pulse-induced genes. In addition, we have examined the effect of caffeine, which M. Brenner and S.D. Thomas (1984, Dev. Biol., 101, 136-146) showed preferentially blocks the cAMP relay system by blocking receptor-mediated activation of adenylate cyclase. We show that in many of the mutants and in an axenic wild-type strain, caffeine causes the induction of pulse-induced gene expression to almost wild-type levels or in some cases to higher than wild-type levels. Our data suggest that caffeine works by activating some step in the signal transduction pathway that must lie downstream from both the receptor and at least one of the G proteins and thus has effects other than simply blocking the receptor-mediated cAMP relay system.