Consistent increases in soybean (Glycine max L.) grain yields over the past decades in Illinois have co‐increased P demand, but—as for many US Midwest states—P fertilizer recommendations are outdated. We evaluated soybean grain yield, P uptake and removal with grain harvest, and P use efficiency in 4 site‐years of field trials on Argiudolls‐Endoaquolls and Fragiudalfs‐Hapludalfs under annual P application treatments of source (monoammonium phosphate [MAP], diammonium phosphate [DAP], triple superphosphate [TSP]), rate (partial vs. full maintenance rate), and timing–placement combination (fall and spring broadcast, spring banding). Substituting ammonium phosphate fertilizers, the most commonly used P fertilizers in the US Midwest, with N‐free TSP supported similar yields and resulted in similar P removal as hypothesized, while avoiding co‐applied N that can be subject to losses primarily via leaching. Soybean yield and grain P removal were unresponsive to rate, timing, and placement even at the partial rate, although banding and spring application may have reduced N loss risk for MAP and DAP. Given the challenges in accurately estimating P removal rate by grain harvest due to the variability in yield and grain P concentrations across years, assessment of P use efficiency should focus on long‐term balance between fertilization (input) and crop removal (output).