Field experiments were conducted during 1989 to 1991 on a loamy sand (Typic ustochrept) soil to study the K nutrition of soybean. Soybean responded significantly up to 50 kg K ha−1 when applied 1/2 of K at planting and 1/2 at flower initiation (two split) or applying 1/3 K at planting, 1/3 at flower initiation and 1/3 at pod development (three splits). Split application was more beneficial than applying full K at time of planting due to higher leaf area index, crop growth rate, chlorophyll content of fresh leaves, K accumulation in soybean and better agronomic and physiological efficiency of applied K. Agronomic efficiency, physiological efficiency and apparent recovery of K reduced as rate of applied K was increased from 50 to 75 kg ha−1. Highest K+ concentration (3.4 % of dry matter) was recorded in 30-day-old plants at 75 kg K ha−1 which depressed progressively with the age of the crop. At maturity, the K concentration of soybean seed varied from 1.5% (unfertilised K plants) to 2.1 % (when 75 kg K ha−1 was applied in three splits).