Abstract: An integral technique to validate Reynolds-averaged Navier–Stokes (RANS) turbulence models is presented. The technique has the advantage of providing a direct connection between wall fluxes and mean flow dynamics, thus providing the necessary means to evaluate if a model correctly predicts the flow physics. In turn, the technique provides needed information critical to the improved development of turbulence models. To assess the value of the technique, it is used to evaluate the performance of two low-Reynolds-number turbulence models against DNS of reciprocating channel flow with heat transfer. The evaluation demonstrates that the integral technique is an improved validation technique compared to standard validation techniques.