The authors examine the empirical evidence on the expectations hypothesis of the term structure of interest rates in the United States, the United Kingdom and Germany using the Campbell-Shiller (1991) regressions and a vector-autoregressive methodology. They argue that anomalies in the U.S. term structure, documented by Campbell and Shiller (1991), may be due to a generalized peso problem in which a high-interest rate regime occurred less frequently in the sample of U.S. data than was rationally anticipated. They formalize this idea as a regime-switching model of short-term interest rates estimated with data from seven countries. Technically, this model extends recent research on regime-switching models with state-dependent transitions to a cross-sectional setting. Use of the small sample distributions generated by the regime-switching model for inference considerably weakens the evidence against the expectations hypothesis, but it remains somewhat implausible that their data generating process produced the U.S. data. However, a model that combines moderate time-variation in term premiums with peso-problem effects is largely consistent with term-structure data from the U.S., UK and Germany.