1. The effect of non-selective (3-isobutyl-1-methylxanthine, IBMX; theophylline) and type IV- or type III/IV-selective (rolipram, RP 73401; zardaverine, tolafentrine) phosphodiesterase (PDE) inhibitors on human eosinophil functions was investigated. 2. For this purpose human eosinophils were purified from blood of healthy donors by a magnetic cell separation (MACS) technique to a purity > or = 99%. From the stimuli investigated (complement C5a; N-formyl-methionyl-leucyl-phenylalanine, fMLP; platelet activating factor, PAF; opsonized zymosan) C5a was selected to test the influence of the above mentioned compounds on secretion of granule constituents (eosinophil cationic protein, ECP; eosinophil-derived neurotoxin, EDN) as well as on formation of reactive oxygen species measured by luminol-enhanced chemiluminescence in intact cells. For comparison, inhibition of PDE IV activity in the cytosol of disrupted cells, which contains about 75% of total PDE IV activity, was determined. 3. Both theophylline and IBMX inhibited the two cell responses with IC50 values which were in the range of their IC50 values obtained for inhibition of PDE IV activity in the cell-free system. The beta 2-adrenoceptor agonist, salbutamol (1 mumol l-1), which by itself did not substantially influence the two cell responses, only marginally improved the potency of theophylline and IBMX in inhibiting ECP/EDN secretion. Only the IC50 value of IBMX for inhibition of chemiluminescence was lowered by about one order of magnitude in the presence of salbutamol. 4. In contrast, none of the selective PDE inhibitors tested substantially inhibited the two cell responses at concentrations up to 10 mumol l-1. This was surprising because all of the compounds investigated inhibited PDE IV activity in the cell-free system with IC50 values which were at least 30 fold lower than the highest concentration of the compounds used with intact cells. In combination with salbutamol, however, both ECP/EDN secretion and chemiluminescence was inhibited by rolipram and zardaverine with IC50 values similar to the IC50 values for inhibition of PDE IV activity. Although RP 73401 and tolafentrine also inhibited both cell responses in the presence of salbutamol, the potency of these two compounds in inhibiting eosinophil function in intact cells was at least two orders of magnitude lower than would have been expected from the inhibition of PDE IV activity in the cell-free system. 5. These results indicate that (i) C5a-stimulated human eosinophils are sensitive to inhibition by then on-selective PDE inhibitors theophylline and IBMX, (ii) the inhibitory effect of these non-selective PDE inhibitors cannot be mimicked by selective PDE IV or PDE III/IV inhibitors although human eosinophils almost exclusively contain PDE IV; (iii) the selective PDE inhibitors need an additional cyclic AMP trigger like a beta 2-adrenoceptor agonist to be effective; but (iv) under the latter conditions inhibition of cell responses in intact cells does not correspond to inhibition of PDE IV activity in the cell-free system.6. We conclude that the non-selective PDE-inhibiting xanthines may inhibit C5a-stimulated human eosinophil responses by other action(s) in addition to PDE IV inhibition, and that inhibition of PDE IV activity in the cell-free system by the selective inhibitors may not generally represent the potency of the compounds in intact cells.