OBJECTIVES: The use of instrumental variable (IV) methods is attractive because, even in the presence of unmeasured confounding, such methods may consistently estimate the average causal effect of an exposure on an outcome. However, for this consistent estimation to be achieved, several strong conditions must hold. We review the definition of an instrumental variable, describe the conditions required to obtain consistent estimates of causal effects, and explore their implications in the context of a recent application of the instrumental variables approach. METHODS: We use two instrumental variables and apply Shea's partial R-square method, the Anderson canonical correlation, and Cragg-Donald tests to check for weak instruments. Hall-Pexie tests are applied to see if any of these instruments are redundant in the analysis. RESULTS: Total 15,956 asthma patients from a private payer data set were examined in this study. We used controller-reliever copay ratio and physician/practice prescribing patterns as an instrument. We demonstrated that the former was a weak and redundant instrument producing inconsistent and inefficient estimates of the effect of treatment. The results were worse than the results from standard regression analysis. CONCLUSIONS: Despite the obvious benefit of instrumental variable models, the method should not be used blindly. Several strong conditions are required for these models to work, and each of them should be tested. Otherwise, the results will be statistically worse than the results achieved by simply using standard ordinary least squares.