Implied by the zero curve
US Treasury forward rate curve
The rate the market is implying for a moment in the future, rather than for a period. Derived from the same fit as the zero curve, not fitted separately.
What a forward rate is
A zero rate answers "what return do I lock in from today out to this maturity". A forward rate answers a different question: "what return is the market implying for a future stretch of time, given those zero rates". If two-year money yields more than one-year money, the extra has to be earned somewhere in the second year, and the forward curve is where that shows up.
This one is instantaneous. It is the rate for an infinitesimal moment at that maturity, not the rate for a year starting there. A "one-year forward one year out" is a period rate and is a different number. Quoting an instantaneous forward as though it were a period rate overstates it wherever the curve is sloping.
It is not a forecast. It is what today's prices imply, and it is the rate at which a borrower could fix future funding now. Those are the same arithmetic and a very different claim.
Why it comes from the zero curve
The zero curve is a Nelson-Siegel-Svensson fit: six parameters describing one smooth curve through the cashflows of every security outstanding that day. The forward rate is that curve's derivative, so it falls out of the same six numbers. There is no separate forward fit to publish, which is why this page carries no fit error of its own and quotes the zero curve's instead.
The curve, September 21, 2026
All families on this day →| Maturity | Forward | Zero |
|---|---|---|
| 1Y | 4.848% | 4.482% |
| 2Y | 4.980% | 4.716% |
| 3Y | 4.879% | 4.789% |
| 5Y | 4.765% | 4.792% |
| 7Y | 4.963% | 4.806% |
| 10Y | 5.485% | 4.931% |
| 15Y | 5.979% | 5.219% |
| 20Y | 5.753% | 5.393% |
Where the forward sits above the zero rate the curve is rising at that maturity; where it sits below, the curve is falling. The two cross wherever the zero curve turns.
Not published beyond 20 years
A derivative amplifies whatever is loose in the thing it is taken from. The zero curve's own error is modest at the long end, and the forward's is not: it reaches about 61 basis points at thirty years against roughly 7 at twenty.
So the last 2 fitted maturities are deliberately withheld here and everywhere else on this site, rather than published with a footnote nobody reads. If you need a thirty-year forward, this data does not support one.