The Bootstrap Statistical method is applied for estimating the accuracy of the convective heat transfer non linear correlation of AL2O3 nanofluid working as cooling fluid. The flow experiment considers laminar and turbulent regimen through an array of aluminum microchannels and millichannels heat sink, taking into account the Volume Fractions, Reynolds, Peclet and Prandtl numbers. The β’s parameters are estimated with nonlinear least square approach. StatGraphics® was used, considering the Gauss-Newton algorithm with Levenberg-Marquardt modifications for global convergence. Correlation for Nusselt number is presented and suggestions for future experimentation are presented in order to improve the accuracy of the regression.