Absolute past geomagnetic field intensity determinations requiring laboratory heating are labourious and the success rate is rather low, mostly because of induced thermochemical magnetic mineral alterations. Archaeomagnetic intensity determinations are mainldy limited to displaced ceramics produced in kilns. In this study, the suitability of an in situ baked structure is investigated. Different magnetic properties of baked material taken from the combustion chamber wall and floor of a Roman pottery kiln, with variable colouring, are examined in dependence on the distance to the combustion chamber. The temperature distribution is re-constructed based on rock magnetic experiments after stepwise heating. The rock magnetic temperature estimates agree fairly well with a mathematical heat conduction model demonstrating the penetration of heat into the combustion chamber wall. The rock magnetic results show that blackish- and greyish-coloured kiln parts, that had been in close contact with the fuel, during ancient kiln operation, are not suitable for intensity determinations. Although sufficiently baked, they strongly alter during laboratory heating and new remanence-carrying minerals are formed. The brownish-coloured material at a distance 65–80 mm away from the combustion chamber seems to be most suitable as its magnetic properties remain nearly unchanged during laboratory heating. Rock magnetic and modelled temperature estimates for this material consistently indicate ancient baking temperatures of about 600◦ C. The model demonstrates that cooling takes longer in the inner parts of the combustion chamber wall. Retarded cooling affects the blocking temperatures and hence the strength of the thermoremanent magnetization. The variability of cooling rates should be taken into account when investigating archaeointensities of specimens cut from large samples, or of samples taken from different parts of a kiln.