Bibliometric studies often exclude documents with little or no scientific content. Yet, identifying and classifying the unscholarly publication is a complex matter, and misclassifications often occur. Reference-based classifications are frequently proposed and implemented in bibliometric studies. Unfortunately, with little support for the actual classifications. In search of valid separation of scholarly/unscholarly publications, this study explores the correlation between number of references and citations received. Data was drawn from Scopus, and two separate analyses were conducted. The first analysis focus on journal articles published in the year 2000 from each of the four major subject areas in Scopus: Health sciences, life sciences, physical sciences and social sciences. The second analysis focus on all journal articles published in either 2000, 2005, 2010, 2015 or 2020. None of the analyses identified a natural cutoff point between scholarly and unscholarly publications. There is no significant point in the distribution that could more clearly distinguish scholarly from unscholarly publications than prior suggestions. In addition, the results of this study demonstrate how bibliometric analyses are impacted by defining scholarly publications on the number of references across different subject areas.