An item response theory model of response stability is developed, based on the local independence principle. The model predicts response changes under repeated administrations of the same instrument using item and examinee parameter estimates as predictors. Real data were used to assess how the model functioned. Results indicated that the model predictions were approximately fulfilled. Limitations of the model and the empirical study are discussed. Index terms: item response models and personality, local independence principle, personality measurement, test-retest stability.