This report analyses justifications gathered in a series of studies which led intellectually normal infants' school and nursery school children and mildly mentally retarded special day school children to acquire conservation understanding. Two hundred and five children were given a pretest of number, mass, length, liquid, and weight conservation. Training groups for non-conservers were formed on the basis of pretest results, mental age and chronological age. Trainees participated in three training sessions given approximately one week apart. All trained children received Learning Set or Verbal Rule training. In a posttest, held approximately one week later, justifications for judgments in the conservation tasks were elicited by asking, "How can you tell?" Judgments and justifications for all conservation tests were compared. It was found that cases of paired correct judgments and incorrect justifications as well as paired incorrect judgments and correct justifications were well below the level that might have occurred by chance. Justifications given were of several different types (identity, reversibility, compensation, addition/subtraction, perception or other) and, except for addition/subtraction justifications, were similar across all groups of children. Trained children, especially the retarded trained children, were more likely to offer multiple justifications. Additional trials on tasks produced more information and improved some children's chances to conserve successfully. (Author/RH)