An inconsistency between the reactivity worth of short-sized samples measured by the critical-water-level (CWL) method and that conventionally analyzed for validating nuclear data and nuclear calculation methods has been known. The present study investigated this inconsistency in terms of a simple theoretical framework and proposed a simple and practical technique for correcting the measured sample reactivity worth without making supplementary experiments. A series of Monte Carlo calculations that simulated typical sample reactivity worth measurement by the CWL method showed that this inconsistency is effectively reduced by the present correction technique.