A comparative study in horses of the pharmacokinetics (PK) and pharmacodynamics (PD) of 2 extensively used nonsteroidal anti-inflammatory drugs (NSAIDs), flunixin (FXN) and ketoprofen (KTP), was carried out applying PK/PD modelling. To evaluate the anti-inflammatory properties of these drugs a model of acute inflammation, comprising surgically implanted subcutaneous tissue cages stimulated by intracaveal injection of carrageenan, was used. FXN elimination half-life (T1/2 beta) in plasma was 3.37 +/- 1.09 h. However, in exudate a much longer T1/2 beta was obtained (15.99 +/- 3.80 h). Apparent volume of distribution (Vdarea) for FXN was 0.317 +/- 0.126 l/kg and body clearance (ClB) was 0.058 +/- 0.004 l/kg/h. KTP displayed enantioselective pharmacokinetics, the S(+) enantiomer being predominant in plasma, exudate and transudate. T1/2 beta values for R(-) and S(+)KTP were, respectively, 1.09 +/- 0.19 h and 1.51 +/- 0.45 h (plasma) and 19.73 +/- 2.72 h and 22.64 +/- 4.34 h (exudate), respectively. R(-)KTP was cleared more rapidly than the S(+) enantiomer. ClB values were 0.277 +/- 0.035 l/kg/h and 0.202 +/- 0.022 l/kg/h, respectively. FXN and KTP pharmacodynamics was evaluated by determining their inhibitory effects on serum thromboxane (Tx)B2, exudate prostaglandin (PG)E2, leukotriene (LT)B4 and beta-glucuronidase (beta-glu) and intradermal bradykinin-induced swelling. Both drugs produced marked inhibition of serum TxB2 synthesis for up to 24 h, with no significant differences between the drugs. FXN was a more potent inhibitor of exudate PGE2, the EC50 for FXN being lower (P < 0.01) than that for KTP (0.019 +/- 0.010 microgram/ml and 0.057 +/- 0.009 microgram/ml, respectively). Neither drug had any effect on exudate LTB4 concentration. Differences between the 2 drugs were observed for the inhibition of beta-glu, the Emax for KTP being higher (P < 0.01) than for FXN. However, no differences were observed in other PD parameters. Both FXN and KTP inhibited bradykinin-induced swelling. Differences between the drugs were obtained for Emax, which was greater for FXN (P < 0.01) than for KTP. Equilibration half-life (T1/2Ke0) also differed, being much longer (P < 0.01) for FXN than for KTP. PK/PD modelling proved to be a useful and novel analytical technique for studying the pharmacodynamics of NSAIDs, with the advantage over classical in vitro methods that it provides data in the whole animal. By quantifying action-concentration interrelationships through PK-PD modelling, it is possible to shed light on molecular mechanisms of drug action, and establish probable differences in mechanisms of action between structurally similar drugs.(ABSTRACT TRUNCATED AT 400 WORDS)