Rationale: Interpretation of longitudinal spirometry in workplace monitoring is often complicated because of unknown quality of spirometry data and the lack of practical and valid statistical tools. We are developing and evaluating the performance of an algorithm for the interpretation of longitudinal spirometry. Method: The algorithm was developed using annual spirometry data from approximately 14,000 workers and evaluated on a subset of 2,577 with >8 years of follow-up. The algorithm is based on monitoring longitudinal data precision to keep it at an acceptable level, and on a referent limit of longitudinal decline which takes into account the data precision and a rate of decline for healthy individuals. Results: The sensitivity of the method to identify individuals with long-term FEV1 decline >90 mL/year increased during the first eight years of follow-up: 4%, 22%, 20%, 29%, 37%, 47%, 51%, and 60%. The likelihood ratio (LR), comparing the likelihood of a positive test when the individual has a long-term decline >90 mL/yr with the likelihood of a positive test when the individual has long-term decline < or = 90 mL/yr, increased during the first eight years of follow-up: l.4, 6.6, 3.9, 9.7, 9.8, 9.7, 11.9, and 12.7. The values were similar for the development of impairment 60% pred. FEV1. Conclusion: The algorithm has a good predictive capacity and allows early identification of procedural errors and early identification of individuals at risk of developing moderate to severe lung function impairment for preventive intervention. The algorithm is now being tested in workplaces using specialized computer software.