One of the fundamental findings in the area of learning and memory is that the repetition of an item increases the probability of its recall. A number of theories have been advanced to explain the effects of repetition on memory. One such theory, encoding variability theory, postulates that changes in the context in which stimuli are presented or changes in the semantic representation of stimuli are directly related to recall. Experiments which have attempted to manipulate semantic variability (e.g., Madigan, 1969) have used no standardized method for selecting stimulus material. Furthermore, those items selected have not been chosen with respect to their degree of semantic variability but some other scale such as frequency of occurrence. In Experiment 1, ratings of semantic meaning change for 147 nouns, biased by different adjectives, were collected from 39 subjects. Two homogenous groups of items then were selected from these normative data and used as list items in a repetition experiment in which subjects were asked to free recall the nouns (Experiment 2). The results from the rating and free recall tasks indicated subjects do make distinctions along a scale of semantic meaning change and the degree of semantic meaning change is predictive of free recall. A model based on encoding variability is presented which assumes that the semantic encoding of an item forms the basis of the memory trace. This model is compared to a similar model of encoding variability recently formulated by Glenberg.