OBJECTIVES: The Regie de l'assurance maladie du Quebec (RAMQ), an administrative database containing medical and pharmaceutical services records, is frequently used in epidemiologic and economic studies. In-hospital length of stay (LOS) and readmissions are important components of such studies and can be estimated through a linkage between RAMQ and MED-ECHO (hospital separation records) databases with high reliability. However, this linkage is cumbersome and time-consuming. These parameters can also be estimated directly from RAMQ medical services database. Unfortunately, the data concerning hospitalizations are not specifically contained in RAMQ and must be deduced from in-hospital medical procedures, claimed by the physician. The objective of this study was to develop and to validate an analytical algorithm to estimate LOS and readmissions in a target population using RAMQ claims database. METHODS: Using the RAMQ medical services database, we identified 1441 patients aged =65 years with hospitalization records, who were treated for chronic obstructive pulmonary disease (COPD) between 1989 and 1996. In-hospital claims were identified using institutional coding. The validation process was then implemented through comparison analyses with a "gold standard", MED-ECHO database, using deterministic linkage method. RESULTS: According to the RAMQ, 1233/1441 patients were admitted at least once during the study period. Compared to MED-ECHO, the sensitivity and specificity of the algorithm identifying the first in-hospital stay (1174 cases) were 97% and 75%, respectively, with ?=0.76. The mean LOS was 12.1 days, with a geometric mean of 6.2 days in RAMQ vs. 13.5 and 6.7 days (p<0.001) in MED-ECHO; the average readmission was 3.7 times vs. 3.8 (p=0.36) for the entire cohort, respectively. CONCLUSIONS: The algorithm is able to predict in-hospital LOS and readmissions with slight underestimation. Substantial time and cost savings can be made through estimation of in-hospital stays using RAMQ database. This algorithm is yet to be validated for other medical conditions.