Purpose:To determine the potential error involved in the interpretation of neutron measurements from medical linear accelerators (linacs) using TLD‐600H and TLD‐700H if standard AmBe and 2⁵2Cf neutron sources are used for calibration without proper inclusion of neutron energy spectrum information.Methods:The Kerma due to neutrons can be calculated from the energy released by various nuclear interactions (elastic and inelastic scatter, (n,α), (n,p), (n,d), (n,t), (n,2n), etc.). The response of each TLD can be considered the sum of the neutron and gamma components; each proportional to the Kerma. Using the difference between the measured TLD responses and the ratio of the calculated Kerma for each material, the neutron component of the response can be calculated. The Monte Carlo code MCNP6 has been used to calculate the neutron energy spectra resulting from photonuclear interactions in a Varian 21EX linac. TLDs have been exposed to the mixed (γ‐n) field produced by a linac and AmBe and 2⁵2Cf standard neutron sources.Results:For dosimetry of neutrons from AmBe or 2⁵2Cf sources, assuming TLD‐700H insensitivity to neutrons will Result in 10% or 20% overestimation of neutron doses respectively.For dosimetry of neutrons produced in a Varian 21EX, applying a calibration factor derived from a standard AmBe or 2⁵2Cf source will Result in an overestimation of neutron fluence, by as much as a factor of 47.The assumption of TLD‐700H insensitivity to neutrons produced by linacs leads to a negligible error due to the extremely high Kerma ratio (600H/700H) of 3000 for the assumed neutron spectrum.Conclusion:Lithium‐enriched TLDs calibrated with AmBe and/or 2⁵2Cf neutron sources are not accurate for use under the neutron energy spectrum produced by a medical linear accelerator.