Soil-test K and estimates of K removal with harvest are tools used to determine K fertilization rates for corn-soybean rotations and other crops.  Incorrect soil-test calibrations and estimates of K removal can have significant economic impacts.  A study based on 20 two-year field trials was conducted from 2003 to 2006 in Iowa to evaluate the effect of K fertilization on corn and soybean grain yield, plant and grain K concentration, and post-harvest soil K measured with ammonium acetate (AA), Mehlich 3 (M3), and sodium-tetraphenylboron (STB) procedures.  Five K treatments (0, 34, 67, 134, and 202 kg K₂O ha^-1) were applied for the first crop and in the second year the plots were subdivided to apply either no K or 134 kg K₂O ha^-1.  A grain yield response was observed in most years and locations when initial AA or M3 K was at levels at which K fertilization is recommended in Iowa (<171 mg K kg^-1).  Plant and grain K concentrations were increased by increasing K rates at most sites, but sometimes K rate effects on post-harvest AA and M3 K were much less than expected.  Measurement of post-harvest non-exchangeable soil K with STB partly explained the results.  Lower than expected K fertilization effects on STK build-up often observed in the Midwest could be explained by too low estimates of K removal and retention in non-exchangeable forms.