We have studied the effects of the internal bremsstrahlung of tritium β-decay, sample surface roughness and tritium content as well on the tritium β-decay induced X-ray spectroscopy (BIXS) for tritium measurements based on Monte Carlo simulations. The Knipp–Uhlenbeck–Bloch theoretical model was adopted for the internal bremsstrahlung. It is found that the contributions of internal bremsstrahlung of tritium β-decays are approximately 30%, 10%, 4% and 2% of the corresponding external bremsstrahlung for C, Ti, Mo and W materials, respectively, and the sample surface roughness may cause an error of several percent in the BIXS method if the surface roughness of a sample is in a reasonable range. Moreover, the effect of tritium content should be taken into account when a sample contains a large amount of tritium, otherwise, the tritium contents measured by the BIXS method will be underestimated, for example, by up to ∼15% for Ti metal.