In this paper, the values of average Nusselt number and heat transfer coefficient ratio for natural convection of silica/water nanofluids in square and triangular cavities are estimated using theoretical correlations. The correlations that are used to calculate the heat transfer characteristics of enclosures are a function of thermophysical properties of nanofluids. Many studies demonstrated that using classic thermophysical models to determine the properties such as thermal conductivity, viscosity, and density sometimes may lead to substantial errors. To avoid such errors, it is helpful to determine the thermophysical properties through experiments. For this purpose, first SiO2 nanoparticles with a diameter of 7 nm are suspended into water to prepare stable nanofluids. Next, thermal conductivity, viscosity, and density of nanofluids with volume fractions of 0.5%, 1%, and 2% are measured at a temperature range of 25-60 degrees C. Measured data are compared with the outcomes of theoretical models. The comparison of experimental data and classic models shows a considerable difference between the outputs particularly at high volume fractions of nanoparticles. Moreover, theoretical models predict a descending trend for variations of thermal conductivity ratio with temperature, but, on the contrary, experimental data unveil an ascending tendency. Finally, Nusselt number and heat transfer coefficient ratio for various cavities including square cavity in horizontal and inclined (angle of 45) positions and the right triangular enclosure are predicted at Rayleigh numbers of 10(5) and 10(6). The findings indicate that even without having measured data of thermophysical properties, the average Nusselt number could be estimated with the same trend and maximum difference of 4.5%. However, the results of heat transfer coefficient ratio reveal the significant of using experimental data for properties to find the optimum working fluid for application in cavities since the theoretical models and experiment-based model may give different trends for heat transfer coefficient in enclosures. (C) 2016 Elsevier Ltd. All rights reserved.