Prescriptive and descriptive grammars of Hungarian frequently discuss variation in the choice of relative pronoun in Hungarian. This paper presents quantitative data about some long standing questions raised by prescriptivists, and a phenomenon that has not been explored: the lack of agreement between the pronoun and the antecedent in Colloquial Hungarian. The study presented here is based on the Budapest Sociolinguistic Interview (Varadi 1998). This corpus consists of 50 sociolinguistic interviews, conducted in 1987 and 1988, totaling approximately 240,000 words. A database was created comprising all nominal relative pronouns from the corpus (N=1714), coded for relevant semantic, syntactic and morphological variables. Two issues, the loss of the relative pronoun amely for specific antecedents, and the spread of the selective relative pronoun amelyik was confirmed. It is also a clear pattern in the corpus, that the plural form of the most frequent relative pronoun ami is avoided, while number agreement is intact everywhere else in Hungarian.