Organic vapor analysis with microsensor arrays relies principally on two output parameters: the response pattern, which provides qualitative information, and the response sensitivity, which determines the limit of detection (LOD). The latter is used to define the operating limit in the low-concentration range, under the implicit assumption that, if a vapor can be detected, it can be identified and differentiated from other vapors on the basis of its response pattern. In this study, the performance of an array of four polymer-coated surface acoustic wave vapor sensors was explored using calibrated response data from 16 solvent vapors in Monte Carlo simulations coupled with pattern recognition analysis. The statistical modeling revealed that the ability to recognize a vapor from its response pattern decreases with decreasing vapor concentration, as expected, but also that the concentration at which errors in vapor recognition become excessive is well above the calculated LOD in most cases, despite the LOD being based on the least sensitive sensor in the array. These results suggest the adoption of a limit of recognition (LOR), defined as the concentration below which a vapor can no longer be reliably recognized from its response pattern, as an additional criterion for evaluating the performance of multisensor arrays. A generalized method for estimating the LOR is presented, as well as a means for improving the LOR via residual error analysis.