The Quotient Method (QM), a laboratory-based risk assessment methodology used by the Environmental Protection Agency to evaluate pesticides for registration and use, has not been thoroughly field-tested and its performance has not always been reliable. My objective was to determine if variation in vegetation structure or diet of exposed animals could result in adverse ecological effects that were not predicted by the QM. In April and early May 1993, I established populations of herbivorous gray-tailed voles (Microtus canicaudus) and omnivorous deer mice (Peromyscus maniculatus) in 24 0.2-ha enclosures planted with alfalfa (Medicago sativa). Alfalfa in 12 enclosures was mowed on 22 June to reduce vegetation height. Small mammal populations were monitored by live trapping from May through August 1993. On 14 July, an organophosphorus insecticide, azinphos-methyl, was applied at 0, 0.88, and 3.61 kg/ha. Insecticide residues were measured on canopy-level spray cards, soil samples, and alfalfa. I compared the observed residue concentrations with predictions based on the nomogram used to estimate exposure for QM risk assessments. I also compared QM predictions of risk with observed effects on population size and growth, survival, reproductive activity, recruitment, body growth, movements, and diet of the small mammals. Much of the insecticide reached ground level in mowed enclosures, but dense alfalfa intercepted most of the spray in unmowed enclosures. The mean half-life of azinphos-methyl on alfalfa was 3.4 days and was not affected by mowing. Mean residue concentrations on mowed alfalfa and the top 15 cm of unmowed alfalfa were underestimated by the QM exposure nomogram. Therefore, pesticides may pose greater risk to organisms inhabiting sparse vegetation or the tops of plants than predicted by the QM. Treatment with azinphos-methyl at 3.61 kg/ha caused severe effects in both mowed and unmowed enclosures on population size and growth, survival, recruitment, and body growth of voles. Effects of azinphos-methyl on vole recruitment and body growth and on survival of female voles were greater in mowed than in unmowed enclosures. However, I did not find that population-level responses of voles to the chemical differed between mowing treatments. Most effects on voles were of short duration (<27 days) but vole densities in 3.61 kg/ha enclosures remained depressed>6 weeks after spraying. The 3.61 kg/ha application rate resulted in a 42% decrease in deer mouse densities in mowed enclosures during the week of spraying, but the insecticide had no adverse effects on deer mice in unmowed enclosures. In addition, the insecticide may have reduced recruitment of deer mice in mowed enclosures. Analysis of deer mouse feces indicated that consumption of arthropods increased in insecticide-treated enclosures just after spraying occurred. Survival, reproductive activity, body growth, and movements of deer mice were highly variable and not significantly affected by azinphos-methyl. Mowing resulted in greater residue concentrations than predicted and, consequently, the insecticide adversely affected voles and deer mice in mowed enclosures at application rates characterized as low risk by the QM. However, food aversion or selective feeding on alfalfa tops may have resulted in similar exposure of voles to the 3.61 kg/ha treatment in mowed and unmowed enclosures. I did not find that insectivorous feeding behavior of deer mice made them more susceptible than predicted. Although residue concentrations on alfalfa did not follow predictions, the gross pattern of effects on small mammals was consistent with QM risk characterization. However, the QM may underestimate exposure and risk when pesticides are sprayed on sparse vegetation.