The U.S. Department of the Treasury publishes Daily Treasury Par Yield Curve Rates on its Interest Rate Statistics page at home.treasury.gov. This dataset shows the relationship between a Treasury security's yield — the annual return an investor receives relative to its price — and its time to maturity, based on closing market bid prices. It is the foundational reference that bond traders, risk managers, and economists rely on to track the government's borrowing costs across different time horizons.

For those who need historical data, Treasury maintains a Daily Treasury Rate Archives page at home.treasury.gov. Users can select "Daily Treasury Par Yield Curve Rates" as a specific data type, alongside two others: Daily Treasury Bill Rates and Daily Treasury Long-Term Rates. The Long-Term Rates series is built as arithmetic averages rather than direct par yields — a distinction that matters when the data is used to benchmark other financial instruments or to set discount-rate assumptions for valuing future cash flows.

Treasury also publishes a separate "Daily Treasury Real Long-Term Rates" series, distinct from the nominal Long-Term Rates (home.treasury.gov; Treasury press release). The real rates series factors in breakeven inflation — the rate of inflation that the market implies by comparing regular Treasury bonds to Treasury Inflation-Protected Securities (TIPS) — giving users an inflation-adjusted counterpart to the nominal averages. Mixing the two without noting the difference in how each is constructed is a common error in downstream analysis.

The Treasury Yield Curve Methodology page, last published February 18, 2025, at home.treasury.gov, documents the technical approach Treasury uses to derive the par yield curve from observed market data. For anyone building or validating curve-fitting models against Treasury's outputs, this page is the primary reference for understanding how the department handles interpolation (estimating values between known data points), smoothing, and the selection of on-the-run versus off-the-run issues — that is, the most recently issued securities at a given maturity versus older ones.

One structural discontinuity in the dataset deserves particular attention. Treasury stopped publishing the 30-year constant maturity series on February 18, 2002, and did not resume it until February 9, 2006 (home.treasury.gov; FRED; Federal Reserve H.15). That gap means any historical time-series analysis spanning the early 2000s will encounter a structural break at the long end of the curve. Models trained on or backtested against constant maturity data from that period need to account for the missing 30-year point, whether through interpolation from adjacent maturities or by flagging the interval as incomplete.

The Treasury Constant Maturity Rates dataset, also available through the Office of Financial Research, provides a daily view of the yield curve's shape across the full maturity spectrum. For those building relative-value frameworks or monitoring curve inversions — when short-term rates exceed long-term rates, a widely watched recession signal — the constant maturity construction allows consistent point-to-point comparison across dates, even as the underlying on-the-run securities roll over time.

The broader context here is that these datasets are not merely reference tables. They serve as the discount-rate backbone for valuing everything from agency mortgage-backed securities to corporate bonds, and they feed directly into the Fed's H.15 statistical release that anchors countless financial contracts and regulatory capital calculations. The distinction between par yields, arithmetic averages, and real rates is not academic. A model that benchmarks against the Daily Treasury Long-Term Rates when it should be using par yields from the curve will systematically misprice duration exposure — that is, sensitivity to interest-rate changes — at the long end.

The 30-year gap from 2002 to 2006 is the kind of structural break that can silently corrupt a backtest. Anyone running multi-decade regression analysis on the term premium (the extra yield investors demand for holding longer-term bonds) or the slope of the curve should verify that their data provider handled the discontinuity explicitly rather than papering over it with interpolation. The resumption date in February 2006 means the series now has roughly two decades of uninterrupted data, but that four-year hole remains a live issue for long-horizon work.

One practical point for those who pull this data programmatically: the distinction between the par yield curve rates and the long-term average rates is not always clearly labeled in third-party redistributions. Treasury's own pages maintain the separation, but Bloomberg, FRED, and other aggregators may present the series with naming conventions that obscure which construction methodology applies. Verifying the source series against Treasury's published methodology page eliminates that ambiguity.