Development of a hybrid approach that attempts to merge LES and RANS methodologies with a view to solving practical problems is explored. The one equation LES model [6, 9] and a length scale RANS model [7, 13, 16] are examined. A priori testing of the LES model using DNS data for isotropic turbulence shows that the LES model correlates very well with the DNS. Examination of the RANS model shows that in order to merge the two models, a variable grid dependent coefficient has to be used. Using this coefficient is seen to mimic the behavior of the eddy viscosity as obtained from a filtered DNS flowfield. The correlations with the subgrid eddy viscosity at fine resolution limits show that the LES model behavior is recovered in the fine mesh limit.