Abstract Maize ( Zea mays L.) is one of the most susceptible crops to zinc (Zn) deficiency. However, in much of the world, soil Zn tests are poorly calibrated, and thus determining a critical soil test, level for Zn is challenging. The objectives of this study were to: (i) produce a field calibration of the Zn Mehlich‐3 (M3‐Zn) method for predicting maize grain yield response to Zn fertilizer application, (ii) compare the capacity of DTPA extractable Zn (DTPA‐Zn) with M3‐Zn for predicting the response of maize yield to Zn fertilizer, (iii) determine whether inclusion of soil pH, organic matter (SOM) and extractable phosphorus in a M3‐Zn model improves its predictive capacity and (iv) evaluate an economic approach for determining soil Zn thresholds. We conducted 55 field experiments covering a wide range of edaphic and climatic conditions. Maize responded to Zn fertilizer in 29% of the trials. The capacity of M3‐Zn and DTPA‐Zn to predict relative yield of maize was similar. Inclusion of other soil variables (pH, extractable phosphorus and SOM) did not or only slightly improve the prediction of M3‐Zn. Based on the relationship between M3‐Zn and DTPA‐Zn ( R 2 = 0.89), one test can be predicted from the other without affecting the calibration. The M3‐Zn 'economic threshold' ranged from 0.98 to 2.79 mg kg −1 , while for DTPA‐Zn it varied from 0.41 to 1.61 mg kg −1 . The broad range of economic thresholds shows that differences in maize productivity and grain price between regions and seasons make establishing a single Zn threshold for all economic‐productive situations inappropriate.