Depression is a latent characteristic that is measured through self-reported or clinician-mediated instruments such as scales and inventories. The precision of depression estimates largely depends on the validity of the items used and on the truthfulness of people responding to these items. The existing methodology in instrumentation based on a factor-analytic approach has limited applicability, especially in the detection of sources of measurement error in itemand person-level analyses. While there are probabilistic approaches such as the use of Item Response Theory and the Rasch model in validating instruments, there are no definite guidelines on the sequence of steps to follow. This study explored the suitability of the Rasch model in assessing and streamlining the University Student Depression Inventory (USDI) using a sequential strategy based on the item response model assumptions, which involves fitting the data to the model through the elimination of misfits, analyzing retained items, and constructing measures. The strategy was applied to two sets of survey data collected from the same population of college students enrolled in a Philippine university but in different semesters. Results showed that the Rasch procedure was able to detect misfit items and persons, which guided decisions regarding the removal of problematic items and persons while preserving the reliability of the original scale. The methodology used was found to be replicable, as the analyses for the two datasets yielded comparable results in terms of number of items retained, item estimates and severity ordering, and distribution of student depression measures.