Theoretical considerations in Dybvig, Ingersoll, and Ross (1996) lead the authors to conclude that long forward and zero-coupon rates can never fall. We examine this conjecture empirically using monthly U.S. Treasury STRIPS data over the period 1990-2000. Based on the Cox, Ingersoll, and Ross (1985) term structure model and a constant-drift adaptation of that model, we find that, contrary to predictions, implied long maturity zero-coupon rates did fall substantially during the last half of this period.