In Brazil soybean is planted in different regions with very different rainfall patterns. Thus, this study aimed to study the water stress caused by both deficit such as excess. The experiment was conducted at the Biosystems Engineering Department-ESALQ/USP in Piracicaba – SP, December 2011 to March 2012. A greenhouse was built, equipped with 44 drainage lysimeters of 1.3m x 1.1m and 0.75m of deep. The experiment was made in randomized block design with four replications. The treatments consisted of three irrigation depth: Deficit with 50% of Evapotranspiration (ETc); Excess ( 150% of ETc) and Full Irrigation (100% ETc) ranging in 11 different forms of occurrence. The crop growth stage was divided into the following subperiods: S1 (All of Vegetative Stage), S2 (Beginning Bloom until Full Pod); S3 (Beginning Seed until Full Seed); S4 (Beginning Maturity until Full Maturity). All treatments received full irrigation, except when they were in the period of application of the respective treatment. The treatments were: deficit all the time, Deficit just in S1, Deficit just in S2, Deficit just in S3, Deficit just in S4, Excess all the time, Excess just in S1, Excess just in S2, Excess just in S3, Excess just in S4, Full Irrigation. Excess was not detrimental to any of the variables. However, the deficit applied during grain filling resulted in significant productivity losses.