Clast shape preferred orientation or anisotropy of magnetic susceptibility (AMS) data are increasingly used to constrain the source location of ash flow tuffs. Little attention has been paid to processing of these directional data for the accurate and objective identification of the sources. This topic is the focus of this paper, as applied to locating the source of the ~5 Ma Kizilkaya ignimbrite, central Turkey. First, geological criteria allowing a rough identification of source vent location are discussed. Second, AMS data are processed to more accurately locate the source of this widespread deposit. Hillhouse and Wells <1991> present a method which consists of determining the areal overlapping density function of the AMS error fans (defined by the Kmax confidence ellipse). Applications to the Kizilkaya ignimbrite show various possible source positions. However, a unique anomaly occurs in every model, suggesting a source in the Derinkuyu basin. Another method is proposed here and is named the average radial angular difference (ARAD) function. It consists of calculating the average angular difference of lines that join a given source point to the sampling sites, with the lineations corresponding to the AMS data interpreted in terms of kinematic axes. Application with the data obtained at the 44 AMS sampling sites shows that the function decreases to <35¿ in the Derinkuyu basin. If the analysis is narrowed to the presumed 12 most proximal sites the function provides values of <20¿ in the central part of the Derinkuyu basin. The identified source area occurs near the source zone of older Miocene ignimbrites and coincides with a gravity anomaly and with a basaltic shadow zone. ¿ 2000 American Geophysical Union