A sliding mode control (SMC) based on the K observer for a three tank non-interacting system is designed and evaluated in this paper. This system is important in modern process control since it can potentially improve product quality and enhance economic benefits. The SMC controller is designed to ensure that the actual liquid level tracks the desired in a finite time. The K observer is used to estimate the speed and acceleration signal, and realize sliding mode control only by using the position signal. The Lyapunov theory is chosen to prove the stability of the control system. The simulation results of the proposed method in MATLAB/Simulink were compared to an optimal controller tuning of a PI controller based on Skogestad’s tuning method, Cohen coon based tuning and Internal Model Control, an optimal PIDA controller with model uncertainty by Cuckoo Search, a fuzzy logic controller, a PID-Fuzzy logic controller. The comparison results showed that the proposed controller was more effective when the rising time reached 2.1521s, the percent of overshoot was 0%, the steady state error converged to zero, the settling time was 3.8577s.