Accurate measurements of the ion and electron temperatures in tokamaks are essential for understanding the heating and transport properties of the plasma. While electron temperature measurements are readily available in most devices (e.g. with electron cyclotron emission or Thomson scattering diagnostics), ion temperature estimates usually require more sophisticated procedures such as charge-exchange recombination spectroscopy [1] or crystal spectroscopy [2]. In the absence of such dedicated ion temperature measurements, a technique often adopted in tokamak plasmas is to estimate the ion temperature by reverse engineering of the total neutron yield, assuming Maxwellian distributions for the ions. In the absence of external ion heating, this procedure is satisfactory since it mainly depends on the nuclear reaction cross sections and the bulk ion density, usually inferred from a combination of the electron density and the plasma dilution measurements. On the other hand, when Ion-Cyclotron Resonance Heating (ICRH) or Neutral-Beam Injection (NBI) is applied, the ion distribution functions are distorted and the Maxwellian representation used for the ion temperature estimate is not necessarily justified, typically leading to an overestimate of the thermal component of the ion temperature. By using a coupled wave / Fokker-Planck numerical solver to calculate the actual ion distribution functions under the influence of external heating and the resulting neutron yield, the fast and thermal components of the bulk ion distributions can be disentangled and a ‘thermal-only’ ion temperature consistent with the neutron measurements can be extracted. An example of this procedure for the WEST tokamak with ICRF heating will be presented and the differences between the results obtained with respect to the simplified Maxwellian-based procedure will be discussed.