This report shows that generation of hydrogen peroxide (H2O2) by human polymorphonuclear leukocytes (PMNs) activated with tumor promoters of varying potency as first and second stage promoters correlates well with activities of these promoters in vivo. Those tested were 12-O-tetradecanoylphorbol-13-acetate (TPA), a complete promoter, 12-O-retinoylphorbol-13-acetate (RPA), a synthetic TPA derivative almost devoid of first stage activity in some strains of mice, and mezerein (Mez), a potent second stage and much weaker first stage promoter. Mez-stimulated PMNs produced up to four times less H2O2, whereas RPA-stimulated PMNs produced up to 10 times less H2O2 than TPA-activated cells when used at concentrations between 0.5 and 15 nM to activate 7.5-8.5 X 10(4) PMNs/ml. Phorbol, a non-promoter, was totally inactive in this assay. Furthermore, the tumor promoter-activated PMNs caused formation of 5-hydroxymethyl-2'-deoxyuridine (HMdU) and thymidine glycol (dTG) in DNA co-incubated with those cells. The amounts of modified thymidines formed, particularly of HMdU, correlated well with first stage tumor promoting efficacy and with the amount of H2O2 that was generated by promoter activated PMNs. In comparison with TPA, Mez- or RPA-stimulated PMNs induced formation of 25 or 70% less H2O2 and 30 or 75% less HMdU, respectively, under conditions favoring HMdU formation. Thus, formation of either H2O2 by tumor promoter-stimulated phagocytes or HMdU in DNA exposed to those activated cells may serve as a measure of potency as a first stage tumor promoter. Formation of modified bases such as HMdU in DNA might constitute the genetic change imparted by the first stage tumor promoters.