In this paper, we analyse the relationship between use and perception of five Catalan variants of the variable "subject coreferential with an antecedent." This will be done by examining the results of a cloze test and three perception surveys answered in 1997 by 26 respondents. Two different correct constructions were then presented as correct answers in the exercise, which were also evaluated in terms of to what extent the subjects used them and how they felt about their grammaticality and their stylistic distribution. Three more variants were also evaluated in the same surveys. The result is that there is general agreement on the stylistic distribution of three variants, ranging from colloquial to formal situations, regardless of the extent the five variants are used.