The Significance Of Unknown Input State Estimators In Advanced Control Engineering Is Immense. In Past Three Decades, Unknown Input State Estimators Have Become A Multi-Purpose Diagnostic Tool In The Field Of Fault Diagnosis. In This Thesis, A Number Of New Kinds Of Unknown Input State Estimators For Both Linear And Nonlinear Systems Are Designed. These Estimators Are The Following: An Unknown Input Kalman Filter For Linear Noisy System, An Unknown Input High Gain Observer For A Linear Uncertain System, A Robust Unknown Input Observer For A Linear System Having Both Noise And Uncertainties, An Unknown Input Nonlinear Observer For An Ideal Nonlinear System And A Robust Unknown Input Nonlinear Observer For Nonlinear Systems With Both Noise And Uncertainties. These Estimators May Be Useful In Robust Control As Well As In Fault Diagnosis. As A Possible Use Of These Estimators, A Component Fault Detection And Isolation Technique Is Developed Using Unknown Input Observers. The Fdi Algorithm Is Devised For Single Fault Hypothesis With The Assumptions That Sensors And Actuators Are Fault Free. The Method Of Extending The Fdi Algorithm For Detecting The Faults Of Lumped Parameter Systems Working Under Different Conditions, For Example, The System With Noise Or Uncertainties Or Nonlinearities, Or Combination Of These, Is Discussed. The Efficacies Of The Estimators As Well As The Fdi Technique Are Shown With The Help Of Different Practical Examples Like Single-Link Robot Arm, Half Car Model, Electro-Hydraulic Actuator, Etc.