The paper investigates asymmetric dependence between the Deutsche mark (DM) and yen against the U.S. dollar (USD). It extends copula theory to allow for conditioning variables, enabling flexible modeling of the conditional dependence structure of exchange rates. The study finds evidence that the DM–USD and Yen–USD exchange rates are more correlated during depreciations against the USD than during appreciations. This asymmetry is consistent with asymmetric central bank behavior and portfolio rebalancing effects. The paper uses a theorem by Sklar to decompose the joint distribution into marginal distributions and a copula, which provides a more informative measure of dependence than linear correlation. It applies this framework to model the joint distribution of DM–USD and Yen–USD exchange rates over 1991–2001, finding significant asymmetric dependence. A structural break in the conditional copula is observed following the introduction of the euro in 1999, with a drop in dependence and a shift in the direction of asymmetry. The study considers two copula models: the symmetrized Joe–Clayton copula and the normal copula. The symmetrized Joe–Clayton copula allows for asymmetric dependence and nests symmetric dependence as a special case. The results show that in the pre-euro period, the yen and DM were more dependent during depreciations against the USD than appreciations, consistent with central bank behavior and portfolio rebalancing. In the post-euro period, the asymmetry reversed, with lower tail dependence dominating upper tail dependence, suggesting a shift in central bank priorities. The paper highlights the importance of testing for goodness-of-fit in copula models and emphasizes the flexibility of copula-based approaches in capturing time-varying dependence structures. It also notes that copulas are more informative than linear correlation in describing complex dependence patterns, especially in non-elliptical distributions. The findings contribute to understanding asymmetric exchange rate dependence and have implications for financial modeling, risk management, and policy analysis.The paper investigates asymmetric dependence between the Deutsche mark (DM) and yen against the U.S. dollar (USD). It extends copula theory to allow for conditioning variables, enabling flexible modeling of the conditional dependence structure of exchange rates. The study finds evidence that the DM–USD and Yen–USD exchange rates are more correlated during depreciations against the USD than during appreciations. This asymmetry is consistent with asymmetric central bank behavior and portfolio rebalancing effects. The paper uses a theorem by Sklar to decompose the joint distribution into marginal distributions and a copula, which provides a more informative measure of dependence than linear correlation. It applies this framework to model the joint distribution of DM–USD and Yen–USD exchange rates over 1991–2001, finding significant asymmetric dependence. A structural break in the conditional copula is observed following the introduction of the euro in 1999, with a drop in dependence and a shift in the direction of asymmetry. The study considers two copula models: the symmetrized Joe–Clayton copula and the normal copula. The symmetrized Joe–Clayton copula allows for asymmetric dependence and nests symmetric dependence as a special case. The results show that in the pre-euro period, the yen and DM were more dependent during depreciations against the USD than appreciations, consistent with central bank behavior and portfolio rebalancing. In the post-euro period, the asymmetry reversed, with lower tail dependence dominating upper tail dependence, suggesting a shift in central bank priorities. The paper highlights the importance of testing for goodness-of-fit in copula models and emphasizes the flexibility of copula-based approaches in capturing time-varying dependence structures. It also notes that copulas are more informative than linear correlation in describing complex dependence patterns, especially in non-elliptical distributions. The findings contribute to understanding asymmetric exchange rate dependence and have implications for financial modeling, risk management, and policy analysis.