Thermal fatigue is a safety relevant damage mechanism in pipework of nuclear power plants. A well-known simplified method for the assessment of thermal fatigue due to turbulent mixing is the so-called sinusoidal method. Temperature fluctuations in the fluid are described by a sinusoidally varying signal at the inner wall of the pipe. Because of limited information on the thermal loading conditions, this approach generally leads to overconservative results. In this paper, a new assessment method is presented, which has the potential of reducing the overconservatism of existing procedures. Artificial fluid temperature signals are generated by superposition of harmonic components with different amplitudes and frequencies. The amplitude-frequency spectrum of the components is modelled by a formula obtained from turbulence theory, whereas the phase differences are assumed to be randomly distributed. Lifetime predictions generated with the new simplified method are compared with lifetime predictions based on real fluid temperature signals, measured in an experimental setup of a mixing tee. Also, preliminary steady-state Computational Fluid Dynamics (CFD) calculations of the total power of the fluctuations are presented. The total power is needed as an input parameter for the spectrum formula in a real-life application. Solution of the transport equation for the total power was included in a CFD code and comparisons with experiments were made. The newly developed simplified method for generating the temperature signal is shown to be adequate for the investigated geometry and flow conditions, and demonstrates possibilities of reducing the conservatism of the sinusoidal method. CFD calculations of the total power show promising results, but further work is needed to develop the approach.