Despite the significant potential of nanofluids in energy storage applications, the experimental determination of thermophysical properties is relatively costly and time-consuming. Therefore, the modeling techniques can be used for the accurate estimation of thermo-physical behavior. The predictive models are useful for the understanding of thermo-physical behavior. Due to restrictions on classical models, there is a need to develop more reliable models to simulate the thermophysical behavior of nanofluids. This paper deals with the synthesis and experimental thermal conductivity measurement of the multi-walled carbon nanotubes (MWCNTs)-Kapok seed oil nanofluid. Additionally, two new correlations based on multiple non-linear regression analysis along well as dimensionless analysis are proposed to estimate thermal conductivity with high precision compared to the classical models. Based on the statistical analysis, the prediction accuracy of the proposed model was ranked. Finally, the sensitivity analysis has been carried out in combination with the residual analysis to assure the accuracy of the model parameters of both proposed models and adequacy of estimated values of model parameters, respectively. The results revealed the global minimum values for all parameters at 0 perturbations indicating that the model parameters were estimated with high accuracy and adequacy. © 2019 Elsevier Ltd