A potential nitrification rate test (PNR) was used to identify metal toxicity in field-contaminated soils. The test was applied to metal salt-spiked soils, to 27 uncontaminated soils, and to 15 soils that are contaminated by former metal smelting activities. Four agricultural soils (pH 4.5-6.6) were spiked with various rates of CdCl2 (0-200 mg Cd/kg dry wt) or ZnCl2 (0-3,000 mg Cd/kg dry wt) and were equilibrated more than nine months prior to testing. The soil Zn EC50s of the PNR were between 150 and 350 mg Zn/kg dry weight. No continuous decrease of the nitrification with increasing Cd application was observed. The nitrification rate was reduced by between 50 and 80% at the highest Cd application in all soils. The PNRs of 27 uncontaminated soils varied widely (0-21 mg N/kg/d), but most of this variability is explained by soil pH (R2 = 0.77). The PNRs of the 15 contaminated soils were 0 to 44% of the values predicted for an uncontaminated soil at corresponding pH. Significant toxicity in field-contaminated soils was identified if the PNR was outside the 95% prediction interval of the PNR for an uncontaminated soil at corresponding pH and was found in seven soils. These soils contain 160 to 34,000 mg Zn/kg dry weight and 5 to 104 mg Cd/kg dry weight and had a pH >5.7. No toxicity could be detected below pH 5.6, where even a zero PNR value is within the 95% prediction interval of uncontaminated soils. It is concluded that the nitrification is sensitive to metal stress but that its power as a soil bioassay is low because of the high variability of the endpoint between uncontaminated soils. The ecological significance of the assay is discussed.