Natural killer (NK) cells are important in combating viral infections and cancer. NK cytolytic activity (NKCA) is often depressed during recovery from strenuous exercise. Lymphocyte subset redistribution and/or inhibition of NK cells via soluble mediators, such as prostaglandin (PG) E2 and cortisol, are suggested as mechanisms. Ten untrained (peak O2 consumption = 44.0 +/- 3.5 ml. kg-1. min-1) men completed at 2-wk intervals a resting control session and three randomized double-blind exercise trials after the oral administration of a placebo, the PG inhibitor indomethacin (75 mg/day for 5 days), or naltrexone (reported elsewhere). Circulating CD3(-)CD16(+)/56(+) NK cell counts, PGE2, cortisol, and NKCA were measured before, at 0.5-h intervals during, and at 2 and 24 h after a 2-h bout of cycle ergometer exercise (65% peak O2 consumption). During placebo and indomethacin conditions, exercise induced significant (P < 0.0001) elevations of NKCA (>100%) and circulating NK cell counts (>350%) compared with corresponding control values. With placebo treatment, total NKCA was suppressed (28%; P < 0.05) 2 h after exercise, and a postexercise elevation (36%; P = 0.02) of circulating PGE2 was negatively correlated (r = 0.475, P = 0.03) with K-562 tumor cell lysis. NK counts were unchanged in the postexercise period, but at this stage CD14(+) monocyte numbers were elevated (P < 0.0001). Indomethacin treatment eliminated the postexercise increase in PGE2 concentration and completely reversed the suppression of total and per CD16(+)56(+) NKCA 2 h after exercise. These data support the hypothesis that the postexercise reduction in NKCA reflects changes in circulating PGE2 rather than a differential lymphocyte redistribution.