Our objective was to evaluate the efficacy, the gastroduodenal safety, and the effects on arachidonic acid products of meloxicam, a new acidic enolic non-steroidal anti-inflammatory drug which preferentially inhibits cyclo-oxygenase-2 over cyclo-oxygenase-1, versus piroxicam in patients with osteoarthritis of the knee. Meloxicam 7.5 mg or piroxicam 20 mg daily was administered for 4 weeks in this double-blind parallel-groups randomised study. The efficacy for pain relief of the two tested medications was assessed by means of visual analogue scale and other clinical parameters. Pre- and post-treatment endoscopies were performed, and the findings were scored and recorded. The gastric fluid was aspirated at each time and prostaglandin E2, thromboxane B2 and leukotriene B4 were determined by ELISA. There was no significant difference between the groups regarding the primary efficacy. Changes in endoscopic findings by means of Lanza score showed statistically significant differences between the two treatment groups in favour of meloxicam at all sites--gastric, duodenal and total. Within-group comparisons showed a statistically significant difference (worsening) in gastric and total score with piroxicam, but no significant difference with meloxicam. The frequency of clinically relevant cases (total score >2) also showed a statistically significant worsening in the piroxicam group. The better GI tolerability of meloxicam was also suggested by fewer adverse GI events and no withdrawals due to adverse events compared with piroxicam. The pre-/post-study gastric juice concentration of PGE2, TXB2, and LTB4 in the meloxicam group was 135.2 +/- 85.8/71.2 +/- 32.2, 116.3 +/- 81.7/99.4 +/- 107.5 and 388 +/- 321/223 +/- 98 pg/ml respectively. The pre-/post-study gastric juice concentration of PGE2, TXB2 and LTB4 in the piroxicam group was 105.7 +/- 43.1/68.2 +/- 34.9, 94.0 +/- 50.9/105.9 +/- 121.1 and 625 +/- 1574/828 +/- 1464 pg/ml, respectively. Both meloxicam and piroxicam significantly inhibited gastric PGE2 levels after 4 weeks' treatment; however, there was no difference between these two groups. Neither of these medications had an effect on TXB2. Only meloxicam inhibited LTB4 concentration significantly, and the between-groups difference was significant. Meloxicam 7.5 mg once daily had better gastrointestinal tolerability and an efficacy comparable to that of piroxicam 20 mg over 4 weeks in patients with osteoarthritis of the knee.