Soil testing is used to evaluate soil K supply for crops. The determination of exchangeable K with the ammonium-acetate extractant (AAK) from dried soil samples is the most widely used soil test for K. Mounting evidence indicates this test is unreliable in some soils and (or) conditions. The objective of this study was to assess the impact of sample drying on the effectiveness of the AAK test in Iowa soils by conducting field calibrations of this test based on dried and field-moist samples for corn and soybean crops. Replicated K response trials were conducted at 120 site-years. Potassium was extracted from samples collected from a 0-15 cm depth that were dried at 40°C (AAK-dry) or not dried (AAK-moist). The AAK-dry test results were 0 to 350% higher than the AAK-moist test results and differences were larger at low K values, which explains a poor correlation across all soils. No single soil property (such as texture, cation exchange capacity, or Ca:Mg:K ratios) explained the differences. Cate-Nelson (CN), linear-plateau (LP) and quadratic-plateau (QP) models were used to estimate critical concentrations (CCK) from relationships between grain relative yield (RY) and K test results. Relationships for AAK-moist showed less data scattering and better defined CCK than for AAK-dry for both crops. The CCK defined by the LP model was intermediate between those defined by CN and QP models. The LP-defined CCK levels for corn were 201 mg K kg^-1 for AAK-dry and 76 mg K kg^-1 for AAK-moist. Similarly, CCK levels for soybean were 215 mg K kg^-1 for AAK-dry and 90 mg K kg^-1 for AAK-moist. Although handling of field-moist samples may increase testing costs, in many conditions a field-moist K test is more reliable than the classic dry test for assessing K sufficiency for crops.