In this paper we first separate inventory data into planned and unexpected parts using Durlauf and Maccini (1995) optimal inventory behavior model, then adopting two econometric methods-OLS and GMM-to estimate and to test this proposition, while using Taiwan’s macroeconomic and different classifications of industry stock data. The empirical results, as we expected, show that there is a negative effect of unexpected inventory on stock returns, which is corresponding to the investigation of the market analysts. Nevertheless, the hypothesis of positive the relationship between the planned inventory and stock returns, which I derive from the theoretical model, is empirically insignificant. This insignificant result may be caused by the improper aggregation of inventories.