Lithium fluoride thermoluminescent detectors are commonly used for neutrons measurements. However, the calibration is usually performed with readily available gamma sources in terms of gamma equivalent air kerma instead of the more meaningful neutron fluence. Neutron calibration coefficients in terms of neutron fluence depend strongly on the experimental conditions and can thus usually not be found in literature. Therefore, the goal of this work was to develop and validate a model to calculate such neutron calibration coefficients and to use this model to quantify the main sources of uncertainty. Comparison of the neutron calibration coefficients predicted by the model with experimental data from literature and this work showed reasonable agreement. However, a slight systematic deviation was found, probably caused by the significant uncertainty on the relative luminescence efficiency values predicted by the recently developed Microdosimetric Model for the low energy charged particles released during the 6Li(n,α)3H reaction. Therefore, it was decided to use relative luminescence efficiency values based on the experimental results instead. The model also identified the neutron energy and angular distribution and the luminescence reader light collection geometry as the main sources of uncertainty. Imprecise knowledge of these experimental input parameters can easily lead to uncertainties of the order of tens of percent on the calculated neutron calibration coefficients. The developed model is applicable for neutron energies below 0.01 MeV, but could be extended to higher energies in the future.