A perplexing phenomenon identified in several tissues is the lack of correlation between inhibition of phosphodiesterase 4 (PDE4) and certain functional responses such as smooth muscle relaxation, gastric acid secretion and cAMP accumulation. Interpretation of these data is complicated further by the finding that function correlates with the ability of PDE4 inhibitors to displace [3H]rolipram [4-(3-cyclopentenyloxy-4-methoxyphenyl)-2-pyrrolidone] from a high-affinity site in rat brain that is apparently distinct from the catalytic centre of the enzyme. We have investigated this discrepancy by using guinea pig macrophages as a source of PDE4 and have confirmed that the ability of a limited range of structurally dissimilar PDE inhibitors (Org 20241, nitraquazone and the enantiomers of rolipram and benafentrine) to increase cAMP content did not correlate with their potency as inhibitors of partly purified PDE4, whereas a significant linear and rank order correlation was found when cAMP accumulation was related to the displacement of [3H]R-(-)-rolipram from a specific site identified in macrophage lysates. An explanation for these data emerged from the finding that the IC50 values and rank order of potency of these compounds for inhibition of partly purified PDE4 and the native (membrane-bound) form of the same enzyme were distinct. Similarly, no correlation was found when membrane-bound PDE4 was compared with the same enzyme that had been solubilized with Triton X-100. These unexpected results were attributable to a selective decrease in the potency of those inhibitors [nitraquazone, R-(-)- and S-(+)-rolipram] that interacted preferentially with the rolipram binding site. Indeed, if membrane-bound PDE4 was used as the enzyme preparation, excellent linear and rank order correlations between inhibition of cAMP hydrolysis, displacement of [3H]R-(-)-rolipram and cAMP accumulation were found, which improved further in the presence of the vanadyl (Vo)/2. GSH complex. Moreover, using Vo/2.GSH-treated membranes, the IC50 values of nitraquazone and the enantiomers of rolipram for the inhibition of PDE4 approached their affinity for the rolipram binding site. Collectively, these data suggest that the rolipram binding site and the catalytic domain on CPPDE4 might represent part of the same entity. In addition, these results support the concept that PDE4 can exist in different conformational states [Barnett, Manning, Cieslinski, Burman, Christensen and Torphy (1995) J. Pharmcol. Exp. Ther. 273, 674-679] and provide evidence that the cAMP content in macrophages is regulated primarily by a conformer of PDE4 for which rolipram has nanomolar affinity.