The osmotic coefficient of B-DNA in water may, in dilute solutions, deviate by as much as 100% from predictions based on a simple line-charge "counterion condensation" theory. In contrast, a cell model description of the ionic atmosphere near a cylindrical polyelectrolyte predicts osmotic properties that are in surprisingly good harmony with all available experimental findings over a wide range of DNA concentrations. We argue that the neglect of molecular features, such as finite radius, makes line-charge condensation theory inapplicable at all but impractically low polyelectrolyte concentrations.