We conducted experiments in 2006 and 2007 in south-western Kenya to reduce labour and input cost of soybean production through different planting systems (point-placement was compared to planting in trenches and broadcasting), weeding frequencies (once or twice), and the use of local inputs (at 20 kg P ha(-1)). Di-ammonium phosphate (DAP) was compared to manure, ashes, combinations of those, 1/2 Tithonia 1/2 DAP (TD) and no input. 1/3rd of the labour was saved when placing seeds in trenches as compared to individual holes without losing significant grain yield. Broadcasting requires 1/9th of the planting time while losing only 15% grain. 5% grain was lost by weeding once instead of twice, while saving 36% of the time. The time saved at planting can be more productively invested in collecting inputs. Plots without input yielded 537 kg ha(-1) grain and 642 kg ha(-1) biomass in 2006. Inputs increased both grain and biomass yields significantly with 27-51%, without significant difference between local and mineral inputs (with exception of TD). Using local input is thus a reasonable decision for farmers. Farmers concluded that the experiments had led to options for different types of farmers, with different access to resources, to increase their soybean production. (C) 2010 Elsevier B.V. All rights reserved.