In this paper, we explore the optical contributions of the substrate to the localized surface plasmon resonance (LSPR) spectrum of surface confined Ag nanoparticles produced by nanosphere lithography (NSL). We present optical extinction spectra of Ag nanoparticles fabricated on the following substrates: fused silica, borosilicate optical glass, mica, and SF-10-a high refractive index specialty glass. For all the experiments discussed here, the Ag nanoparticles were approximately 100 nm in in-plane width and 25 nm in out-of-plane height. In a controlled N-2 environment, the wavelength corresponding to the extinction maximum lambda (max), shifts to the red with increasing refractive index of the substrate, n(substrate). The sensitivity factor, Delta lambda (max)/Deltan(substrate), was measured to be 87 nm per refractive index unit (RIU). Experimental extinction spectra were modeled using the discrete dipole approximation (DDA). The DDA theory qualitatively predicts the experimentally observed trend that lambda (max) is linearly dependent on n(substrate); however, the theory overestimates the sensitivity by approximately a factor of 2. The sensitivity of the LSPR to the refractive index of bulk external solvent, n(external), was also examined for each of the four substrates listed above. For all the cases, the change in II,,, in response to bulk external solvent was linearly dependent upon n(external). Values of the sensitivity factors, Delta lambda (max)/Deltan(external), ranged from 206 nm RIU-1 for mica to 258 nm RIU-1 for SF-10, a difference of only 25%. From the results presented here, we conclude that there is no systematic dependence, or at most a weak dependence, which correlates the bulk solvent sensitivity of the LSPR to n(substrate). The DDA theory overestimates the LSPR sensitivity to bulk external environment, but the ratio of solvent to substrate sensitivity factors is correct within experimental uncertainty. This ratio has a value of approximately 2, which indicates that there is greater sensitivity in the optical response to the solvent than to the substrate. This ratio is within 10% of the ratio of areas of the particles that are exposed to solvent and substrate. We suggest that chemical interactions at the interfaces between the nanoparticle and the substrate and/or the nanoparticle and the bulk environment contribute significantly to the observed difference between experimental and theoretical sensitivity factors.