Exports respond unpredictably to a change in real exchange rates, suggests evidence from the 1980s. Recent theoretical work explains this as a consequence of the sunk costs associated with breaking into foreign markets. Sunk costs include the cost of packaging, upgrading product quality, establishing marketing channels, and accumulating information on demand sources. The authors use micro panel data to estimate a dynamic discrete-choice model of participation in export markets, a model derived from the Krugman-Baldwin sunk-cost hysteresis framework. Applying the model to data on manufacturing plants in Colombia (1981-89), they test for the presence of sunk entry costs and quantify the importance of those costs in explaining export patterns. The econometric results reject the hypothesis that sunk costs are zero. The results, which control for both observed and unobserved sources of plant heterogeneity, indicate that prior export market experience has a substantial effect on the probability of exporting, but its effect depreciates fairly quickly. The reentry costs of plants that have been out of the export market for a year are substantially lower than the costs of a first-time exporter. After a year out of the export market, however, the reentry costs are not significantly different from the entry costs. Plant characteristics are also associated with export behavior: large old plants owned by corporations are more likely to export than other plants. Variations in plant-level cost and demand conditions have much less effect on the profitability of exporting than variations in macroeconomic conditions and sunk costs do. It appears especially difficult to break into foreign markets during periods of world recession.