Thermal transpiration in a cylindrical tube is studied by using the Bhatnagar‐Gross‐Krook model and methods previously used by Ferziger and Cercignani and Sernagiotto in solving the Poiseuille flow problem. It appears that recent results of Sone and Yamamoto are in slight error. Finally, in analogy with the recent work of Mason and co‐workers on the dusty gas model, a relation forftr, the translational Eucken factor, is developed.