The aim of this paper is to present a new routine that produces consistent industrial or product classifications over time. Our routine is useful for empirical researchers since many classification systems such as the Harmonized System (HS) are regularly subject to revisions. The application of our routine is likely to improve the reliability of studies about product survival, churning or scope in export markets and increase the efficiency of “within estimators”. The code that is implemented as a Stata command can be easily adapted to other programming (such as R) or data environments. Our routine based on an object-oriented Java program is less prone to coding errors than existing alternative algorithms that rely on long functional procedures instead.