In this work, localized surface plasmon resonance based fiber optic (LSPR-FO) sensor utilizing gold nanoparticles (Au NPs) with different sizes is theoretically constructed and analyzed. Three layers (fiber core, Au NPs, and the Analyte) configuration designed by using the matrix method. Performance parameters such as sensitivity, signal to noise ratio, the figure of merit, and resolution of the sensor are evaluated for each size of Au NPs. As a result, sensitivity and resolution exhibit no change with Au NPs size change while the signal to noise ratio and figure of merit decrease as Au NPs size increase. This work shows that the optimized LSPR-FO sensor may have highly promising applications in chemical sensing.