Major compounds of several commonly used botanicals, including turmeric, have been purported to have anti-inflammatory actions. In order to test the anti-inflammatory activity of compounds isolated from rhizomes of Curcuma longa L. (Zingiberaceae), we have established an in vitro test system. HL-60 cells were differentiated and exposed to lipopolysaccharide (LPS) from Escherichia coli (1 microg/ml) in the presence or absence of botanical compounds for 24 h. Supernatants were collected and analyzed for the production of tumor necrosis factor alpha (TNF-alpha) and prostaglandin E2 (PGE2) using standard ELISA assays. Water-soluble extracts were not cytotoxic and did not exhibit biological activity. Organic extracts of turmeric were cytotoxic only at concentrations above 50 microg/ml. Crude organic extracts of turmeric were capable of inhibiting LPS-induced TNF-alpha (IC50 value = 15.2 microg/ml) and PGE2 (IC50 value = 0.92 microg/ml) production. Purified curcumin was more active than either demethoxy- or bisdemethoxycurcumin. Fractions and subfractions of turmeric extracts collected via preparative HPLC had differing biological activity, ranging from no activity to IC50 values of < 1 microg/ml. For some fractions, subfractionation resulted in a loss of activity, indicating interaction of the compounds within the fraction to produce an anti-inflammatory effect. A combination of several of the fractions that contain the turmeric oils was more effective than the curcuminoids at inhibiting PGE2. While curcumin inhibited COX-2 expression, turmeric oils had no effect on levels of COX-2 mRNA.