Abstract We explore the behavior of the divorce rates in 17 OECD countries for the period 1960–2010. Many studies have found persistence in divorce rates after changes in divorce laws. We formulate a nonlinear model to explain the behavior of divorce rates, over time, after a large change in divorce rate, including changes in divorce laws.We argue that the stationary smooth transition regression (STR) approach is useful in analyzing divorce rates. We reject the persistence result and, in addition, provide evidence on asymmetries in divorce rates. This suggests that one should be careful when interpreting the effects of the changes in divorce rates (especially changes in the divorce and family laws) on divorce rates.