Summer is loosening its grip on the national parks. Take Yellowstone. A decade ago, June, July, and August pulled in 66% of the year's visits; now it is 58%. Almost all of that lost share moved into the shoulder months on either side of summer, with April and May, September and October together climbing from 31% to 38% of the year. And Yellowstone is no outlier. Of the 354 NPS units with complete monthly data from 2012 to 2024, 233 (66%) gained shoulder-season share, and 184 (52%) gained at least a full percentage point. It turns up at the big-name parks and at small historical sites almost nobody writes about. One caveat, stated up front: several flagship parks rolled out timed-entry or vehicle reservations between 2020 and 2024, and those systems cap summer access by design, which pushes visits toward the shoulders. The table below marks them. The pattern holds even with those parks set aside.
Shoulder vs peak shares at 15 illustrative units, ~2014 vs ~2024
Geographically and designation-diverse subset, sorted by largest shoulder-share gain. The full 354-unit distribution is in the downloadable CSV. "Shoulder" = April, May, September, October as a share of annual Total Recreation Visits. "Peak" = June, July, August. Both windows are 4-year averages.
| Park | Shoulder ~2014 | Shoulder ~2024 | Δ shoulder (pp) | Δ peak (pp) | Timed entry / vehicle reservation? | Avg annual visits (~2024) |
|---|---|---|---|---|---|---|
| Glacier Bay NP & PreserveAK · Cruise-dominated (see note) | 27.9% | 40.0% | +12.1 | −12.1 | No | 518,867 |
| Yellowstone NPWY / MT / ID · National Park | 30.9% | 38.0% | +7.1 | −8.1 | No | 4,349,055 |
| Glacier NPMT · National Park | 24.7% | 30.9% | +6.2 | −6.7 | Yes (2021) | 3,033,121 |
| Acadia NPME · National Park | 35.8% | 41.4% | +5.5 | −5.6 | Yes (2020) | 3,970,227 |
| Badlands NPSD · National Park | 25.7% | 31.1% | +5.4 | −5.2 | No | 1,092,920 |
| Grand Teton NPWY · National Park | 28.5% | 33.8% | +5.4 | −6.1 | No | 3,434,195 |
| Bryce Canyon NPUT · National Park | 42.1% | 47.3% | +5.2 | −6.8 | No | 2,354,651 |
| Mammoth Cave NPKY · National Park | 33.7% | 38.1% | +4.4 | −11.0 | No | 645,103 |
| Mesa Verde NPCO · National Park | 34.7% | 38.9% | +4.3 | −5.1 | No | 508,382 |
| Rocky Mountain NPCO · National Park | 31.1% | 35.0% | +3.9 | −6.7 | Yes (2020) | 4,251,364 |
| Mount Rainier NPWA · National Park | 27.7% | 30.7% | +3.0 | −2.5 | Yes (2024) | 1,463,486 |
| Olympic NPWA · National Park | 31.1% | 32.8% | +1.8 | −2.0 | No | 2,962,654 |
| Great Smoky Mountains NPTN / NC · Includes through-traffic | 38.3% | 39.4% | +1.1 | −1.2 | No | 12,041,229 |
| Arches NPUT · National Park | 43.7% | 43.4% | −0.3 | +0.4 | Yes (2022) | 1,535,488 |
| Shenandoah NPVA · National Park | 47.9% | 47.1% | −0.8 | +1.0 | No | 1,650,571 |
Shoulder-share by park, ~2014 vs ~2024 (illustrative subset)
Most park units are shifting toward the shoulder season
The shoulder-season shift is weakest in the South
How this was computed
Source dataset. National Park Service Visitor Use Statistics, 2025 release. The underlying data is the “Total Recreation Visits” (TRV) statistic published per NPS unit per year per month at irma.nps.gov/STATS. We use the full-package monthly file (Main_Data.csv).
Windows compared. Two symmetric four-year windows: ~2014 = average of years 2012, 2013, 2014, 2015. ~2024 = average of years 2021, 2022, 2023, 2024. Four-year averages reduce single-year noise (fire closures, government shutdowns, COVID-2020).
Shoulder, peak, and shares. “Shoulder” months are April, May, September, October. “Peak” months are June, July, August. A park's shoulder-share for a window is total shoulder-month visits in that window divided by total annual visits in that window, times 100.
Inclusion criteria. Every NPS unit with complete 12-month data in all four years of both windows AND average annual visits ≥10,000 (to suppress noise-dominated tiny units). Of 406 reporting units, 354 qualify; all 354 are in the downloadable full-distribution CSV. The 15-unit on-page table is a geographic + designation diverse subset chosen for illustrative purposes, not for the largest movers.
Policy confound, flagged in the table. Several flagship parks adopted timed-entry or vehicle-reservation systems between 2020 and 2024. These mechanically cap summer access and shift visits to shoulder months. The relevant units are Acadia (Cadillac Summit Road, 2020), Rocky Mountain (timed entry, 2020), Glacier (Going-to-the-Sun corridor reservation, 2021), Arches (park-wide timed entry, 2022), and Mount Rainier (Paradise and Sunrise, 2024). Yosemite ran reservations off and on across 2020 to 2024. The shoulder-share shift is visible at parks without timed entry too (Yellowstone, Grand Teton, Bryce Canyon, Mesa Verde, Mammoth Cave, Badlands). The underlying behavior shift is real, but policy is part of the explanation at several flagship units.
Glacier Bay caveat. Glacier Bay NP & Preserve has the largest shoulder-share gain in the illustrative subset (+12.1 pp), but its monthly visitation is dominated by Alaska cruise itineraries. The shift reflects cruise-line scheduling decisions as much as visitor behavior change. Yellowstone is the cleaner headline. Glacier Bay is shown for completeness.
Great Smoky Mountains caveat. GRSM's TRV figure includes substantial commuter and through-traffic via U.S. 441; the within-park visit share interpretation should be read with that in mind. The differential (within-unit, year-over-year) comparison remains valid.
Other limitations. TRV counts recreation visits, not unique visitors. Per-vehicle multipliers and counting methodology vary by park and have been revised at some parks within the window (e.g., Yellowstone's persons-per-vehicle revision); within-park year-over-year comparison is robust to those changes because the same method applies in both windows.
What this finding does, and doesn’t, tell us
What's solid. The rise shows up at about two-thirds of the units we could measure, in every region, and well beyond the famous parks. No single park's operational quirk could produce a pattern that broad.
What the numbers can't settle. At least four forces could be driving this, and visitation counts alone can't tell you how much each one contributes. Climate: warmer Aprils and Octobers stretch the comfortable-weather window. Policy: timed-entry and reservation systems at the big parks cap summer and shunt people into May and September. The pandemic hangover: off-peak travel habits that formed in 2020 and 2021 appear to have stuck. Who's traveling: retirees, remote workers, and travelers without school-age kids can all go in the shoulder weeks, and more of them are.
The case for real behavior change. Six parks make this hard to dismiss as a reservation-system artifact. Yellowstone, Grand Teton, Bryce Canyon, Mesa Verde, Mammoth Cave, and Badlands all gained 4 to 7 points of shoulder share with no timed-entry program in either window. Nobody capped their summers. Visitors moved on their own.
The case for policy. The other side of the ledger: Acadia, Glacier, Rocky Mountain, and Arches all brought in access limits between 2020 and 2022, and all four moved with the trend. Strip out the policy timeline and their shifts look like everyone else's. Leave it in, and some share of their change belongs to the reservation system, not the visitor.
Yellowstone monthly profile, ~2014 vs ~2024
Quotable sentences
Three short sentences ready to quote with attribution to Nicholas Major, National Sites Guide. When you publish online, a link back to nationalsitesguide.com is appreciated. If you need a phone or video interview, reach out via the contact form.
“Two-thirds of the park units I could measure gained shoulder-season share over the decade. This isn't one crowded park or one region. It's the whole system drifting the same way.”Nicholas Major, National Sites Guide
“Yellowstone is the cleanest case I have. There's no reservation system muddying it, and summer still fell from 66 percent of the year's visits to 58. People moved into May, September, and October on their own.”Nicholas Major, National Sites Guide
“At parks like Acadia and Glacier, the timed-entry systems make it genuinely hard to say how much of the shift is the policy and how much is the visitor. At the parks with no such system, there's nothing left to argue about.”Nicholas Major, National Sites Guide
How to cite this analysis
If you reference these numbers in print, broadcast, or online, the following one-line citation is preferred:
Data and chart files
All assets on this page are released under CC BY 4.0. Please credit National Sites Guide when republishing.
Who did this work
Reach the analyst
Reach the analyst through the contact form. Mention “press” in your message and we’ll route it accordingly.
Independent site. Not affiliated with or endorsed by the National Park Service. The underlying data comes from the official NPS Visitor Use Statistics Data Package, 2025 release; editorial analysis and methodology are ours. The NPS Arrowhead and other NPS marks are not used.