Grounded theory is a frequently used approach in nursing research. Over the years the methodology has developed in different directions with ambiguous answers to questions of truth and validity. This ambiguity influences the interpretation of the criteria for quality judgement of grounded theories: fit, work, relevance and modifiability. In particular, the criterion fit seems to be caught in a vacuum between different epistemological and ontological positions. Fit can be interpreted either from a realist or from a nonrealist position but both present problems. A realist position is problematic if it insists on an immutable empirical world and ignores the social and psychological aspects of human life. A nonrealist position can either be argued to rely on hidden realist assumptions and therefore to be inconsistent, or it can be relativistic, opening up the possibility of ‘anything goes’ attitudes in research and solipsistic confirmations of the world view of researchers with little or misleading practical impact. A reconsideration of the realist position is suggested, in which validity is regulated by the social constructed reality ‘as it really is’. From this position fit is a matter of correspondence to facts in social reality. The criteria work, relevance and modifiability are argued to support the fitness of a theory, and to be useful in the broader evaluation of the quality of grounded theories.