In 2016, more than 1060 million people in the world lived in areas that are difficult to access, causing insufficient access to electricity. To overcome the problem, a pico hydro cross-flow turbine is considered to be a proper solution because it has a simple design, has few civil works, is easy to maintain, and can operate at a medium head with high variation of discharge. To increase the turbine's performance, this study aims to find the effect of blade depth on performance and to investigate the possibility of a cross-flow turbine being a reaction turbine. The CFD method was selected because it can represent the flow pattern in a turbine with more detail than other methods. The blade depth variation in this study consists of 0 mm, 3 mm, 6 mm and 9 mm, and the pressure inlet boundary conditions are varied with heads of 2.7 m and 5 m. The model turbulence RNG k-standard has been used to predict turbulent flow. From the 2.7 m head, the average efficiency with the ratio U/V 0.42-0.5 produced by blade depth are: 0 mm is 41.9%, 3 mm is 45.8%, 6 mm is 34.4% and 9 mm is 36.7%. Meanwhile, the variations from the 5 m head are: 0 mm is 49.8%, 3 mm is 57.3%, 6 mm is 53.7% and 9 mm is 49.6%. A two-factor ANOVA without replication was performed to determine the relationship of blade depth to performance, and the results showed there is an effect because the F-critical was higher than F. In addition, the blade does not entirely convert the water's kinetic energy to power. Thus, the reaction turbine concept cannot be used in cross-flow turbines because there is no lift force produced by any blade in the two conditions.