Cholinergic muscarinic receptors, when stimulated by arecoline, can activate cytosolic phospholipase A(2) (cPLA(2)) to release arachidonic acid (AA) from membrane phospholipid. This signal can be imaged in the brain in vivo using quantitative autoradiography following the intravenous injection of radiolabeled AA, as an increment in a regional brain AA incorporation coefficient k*. Arecoline increases k* significantly in brain regions having muscarinic M(1,3,5) receptors in wild-type but not in cyclooxygenase (COX)-2 knockout mice. To further clarify the roles of COX enzymes in the AA signal, in this paper we imaged k* following arecoline (5 mg/kg i.p.) or saline in each of 81 brain regions of unanesthetized rats pretreated 6 h earlier with the non-selective COX inhibitor flurbiprofen (FB, 60 mg/kg s.c.) or with vehicle. Baseline values of k* were unaffected by FB treatment, which however reduced by 80% baseline brain concentrations of prostaglandin E(2) (PGE(2)) and thromboxane B(2) (TXB(2)), eicosanoids preferentially derived from AA via COX-2 and COX-1, respectively. In vehicle-pretreated rats, arecoline increased the brain PGE(2) but not TXB(2) concentration, as well as values for k* in 77 of the 81 brain regions. FB-pretreatment prevented these arecoline-provoked changes. These results and those reported in COX-2 knockout mice suggest that the AA released in brain following muscarinic receptor-mediated activation is lost via COX-2 to PGE(2) but not via COX-1 to TXB(2), and that increments in k* following arecoline largely represent replacement by unesterified plasma AA of this loss.