Pyroclastic flow deposits, known as ash-flow tuffs or ignimbrites, are invaluable materials for paleomagnetic studies, with many applications for geological and tectonic purposes. However, little attention has been paid to evaluating the consistency and reliability of the paleomagnetic data when results are obtained on a single volcanic unit with uneven magnetic mineralogy. In this work we investigate this issue by concentrating on the Kizilkaya ignimbrite, the youngest large-volume unit of the Neogene ignimbrite sequence of the Central Anatolian Volcanic Province in Turkey, bringing evidence of significant magnetic heterogeneities in ignimbrite deposits (magnetic mineralogy, susceptibility, magnetic remanence, coercivity, etc.) and emphasizing the importance of a stratigraphic sampling strategy for this type of volcanic rocks in order to obtain reliable paleomagnetic data. Six sections were sampled at different stratigraphic heights within the devitrified portion of the ignimbrite. Isothermal remanence measurements point to low-Ti titanomagnetite as the main magnetic carrier at all sites; at some sites, the occurrence of oxidized Ti-magnetite and hematite is disclosed. The bulk susceptibility (k(m)) decreases vertically at two out of six sections: its value for the topmost samples is commonly one order of magnitude lower than that of the samples at the base. In most cases, low k(m) values relate to high coercivity of remanence (B-CR) values, which range from 25 to >400 mT, and to low S-ratios (measured at 03 T) between 0.28 and 0.99. These data point to the occurrence of oxidized magnetic phases. We therefore consider the k(m) parameter as a reliable proxy to check the ignimbrite oxidation stage and to detect the presence of oxidized Ti-magnetite and hematite within the deposit. The characteristic remanent magnetization is determined after stepwise thermal and AF demagnetization and clearly isolated by principal component analysis at most sites. For these sites, the site mean paleomagnetic direction is consistent with data from the literature. At a few other sites, the remanence is more complex: the direction moves along a great circle during demagnetization and no stable end-point is reached. The occurrence of oxidized Ti-magnetite or hematite as well as two remanence components with overlapping coercivity and blocking temperature spectra suggest that the Kizilkaya ignimbrite acquired first a thermal remanent magnetization and then, during the final cooling or a short time later, a secondary remanent magnetization component which is interpreted as a CRM acquired during post-emplacement devitrification processes. Notwithstanding the Kizilkaya ignimbrite is a single cooling unit, its magnetic properties suffered substantial variations laterally and vertically within the deposit. The Kizilkaya case shows that thick pyroclastic deposits should be sampled using a stratigraphic approach, at different sites and different stratigraphic heights at each individual sampling location, otherwise, under-sampling may significantly affect the paleomagnetic results. When sampling is performed on a short duration or on very poorly preserved deposits we recommend drilling the lower-central portion in the most strongly welded and devitrified fades. Such sampling strategy avoids complications arising from the potential presence of a pervasive secondary CRM masking the original ChRM. (C) 2017 Elsevier B.V. All rights reserved.