Using a sample of firms that adopted the LIFO method of inventory valuation, this paper uses survival analysis to examine the length of time that elapses between the adoption of LIFO and the initial liquidation of LIFO inventory. Prior research in the area of LIFO liquidation uses cross-sectional or pooled, cross-sectional data to model either the change in the LIFO reserve or the probability of a LIFO liquidation within a specific year. This research uses longitudinal data to examine the effect of tax and non-tax costs on the expected time to liquidation after the adoption of LIFO. The results suggest that length of time is influenced by changes in demand for a firm's products and earnings management. The initial liquidation of LIFO inventory is driven by changes in firm sales and the ability of the income effect of the liquidation to allow a firm to exceed prior year's earnings. Unlike earlier research, tax and liquidity considerations appear to play no role in the decision to liquidate inventory. The results in the paper suggests that modeling the length of time that precedes the occurrence of an event and modeling the occurrence of an event itself are independent research questions and that variables that influence outcome need not necessarily influence duration. Given that a large amount of accounting research uses choice-based models to examine the occurrence of events, the results suggest that placing choice-based research within a survival analysis framework has the potential to yield new insights.