摘要:The present paper addresses some specific methodological issues in using toponymy as historical evidence of the ethnic composition of an area in the past. In the Hungarian context, there are two kinds of onomastic data that can be used for this purpose: a) place names borrowed from a language in contact with Hungarian, and b) place names (mainly settlement names) derived from ethnonyms. The authors focus on the second kind of toponymic data survived in medieval charters and historical narratives written mostly in the Latin (and, occasionally, Greek) language. Starting from the structural classification of such material and its quantitative characteristics, the paper consequently demonstrates some patterns of the chronological and territorial distribution of toponymic attestations. Although the research confirms that settlement names derived from ethnonyms may be useful for the reconstruction of the ethnic composition of the Carpathian Basin in the medieval period, such onomastic evidence must be applied with discretion, taking into account the functional specificity of ethnonyms in the studied period as well as the nature of written sources they are retrieved from. The study of Old Hungarian settlement names with ethnonymic elements shows that the latter do not necessarily refer to the ethnicity of the inhabitants, but to the ethnicity of the owner (or another important inhabitant) of the settlement. The authors point out to difficulty of differentiating secondary de-anthroponymic toponyms from actual de-ethnonymic settlement names in medieval sources, warning against straightforward historical interpretation of such onomastic material.