The objective of this paper is twofold. First, against the background of existing empirical literature on the duration of trade which has found that international trade is often of strikingly short duration, we aim to establish whether or not EU imports from the rest of the world also are short-lived. Second, since there is at this point no clear commonly accepted theoretical explanation for these short trade durations, we seek to provide a thorough empirical description and analysis of the phenomenon, with the intention of thereby facilitating theoretical developments on the subject. We employ a rich data set of detailed imports to the EU15 countries from 140 exporters, covering the time period 1962-2006. Using these data, we begin by conducting a thorough descriptive analysis of the duration of EU imports. Thereafter, we perform a regression analysis using discrete-time duration models with proper controls for unobserved heterogeneity. We draw the conclusion that EU imports are indeed very short-lived – in fact, possibly more so than, for example, US imports. The median duration of EU imports is for example merely one year, and almost 60 percent of all spells cease during the first year of service. Among our empirical findings are (i) that the duration of trade remains stable across the long time period that we study; (ii) that short trade durations are the result of at least two processes: countries shifting between different suppliers but continuing to import a given product, and countries ceasing to import the product altogether; and (iii) that countries with a diversified export structure also will tend to have more long-lived export flows. In our formal regression, we are also able to find a set of explanatory variables that have statistically significant effects on the probability that trade flows die.