Previous research has found that evaluations of attribute values (e.g., temperatures, prices, portion sizes, body sizes, sweetness, pleasure, pain, happiness, etc.) depend upon their relationship to default, normal, or average values (reference-point effects; see Helson, 1964; Kahneman & Tversky, 1979) and the type of distribution from which attribute values are drawn (e.g., positively vs. negatively skewed distributions; distribution-density effects; see Parducci, 1995). In this talk, I present the possibility that both reference-point effects and distribution-density effects are created by verbal comparisons (e.g., .I am happier today than I was yesterday.). I describe a model of attribute evaluation in which comparisons mediate the biases in evaluation functions created by reference points and distribution density. I present preliminary evidence in favor of this model and outline continuing research here at DePaul University.