Sauna vs Ice Bath Running Costs: 50-State Data Study (2026)

We have an 8ft Canadian cedar barrel sauna in the backyard with a 9 kW heater, used two or three times a week, and beside it a cold plunge that is an ice bath with no chiller. About three years of that. The question we could never answer properly for readers was the simple one: what does the pair of them cost to run?

Nobody publishes a good number. What circulates is "about a dollar a session" with no working shown, or a manufacturer figure with no stated usage pattern behind it. So we built the number instead, for all 50 states and DC, out of published federal data — and we are publishing the working, the assumptions and the raw files along with it.

One thing to be straight about before the tables: our sauna has never been metered. Everything below is a transparent model built on the heater's nameplate rating, published duty-cycle guidance and state electricity prices. No figure on this page is a meter reading, a test result or anything we observed. If you cite it, cite it as a model — that is what makes it checkable.

The headline numbers

At the EIA US average residential electricity price of 18.11 ¢/kWh (year-to-date through May 2026):

The finding worth quoting: at US average prices the ice bath costs roughly four times as much to run per year as the sauna — about $859 against about $207 — despite using no electricity at all.

That ratio, 4.16×, has two usage bases attached to it and they are not the same: 130 sauna sessions a year (2.5 a week) against 52 plunge fills a year (one a week). Quote it with those bases or it stops meaning anything. It is also a national-average figure — per state it runs from 0.56× in New York up to 10.76× in Florida, and nominally lower still in the ten states where the model returns zero ice and the ratio stops being a real number. The table below shows why.

The full dataset: all 50 states + DC

Central scenario throughout. Sauna: 9 kW heater, 45-minute heat-up, 45-minute session, 30% maintenance duty cycle, 2.5 sessions a week — 1,141 kWh a year in every state, so the entire spread is electricity price. Plunge: 100 US gallons, 50 °F target, one fresh fill a week, ice at $5.00 per 20 lb bag, water at EPA's national combined rate.

† — in these ten jurisdictions (AK, ID, ME, MN, MT, ND, NH, VT, WI, WY) the estimated annual mean tap temperature is already at or below the 50 °F target, so the model returns zero ice and the annual figure shown is water only. That does not mean an ice bath is free in Alaska — summer tap water in all ten runs well above 50 °F and will need ice. Read the artifact section below before quoting any of these rows, and treat their ratio column as an artifact of the annual mean rather than a real number.

The final column is arithmetic on the two columns before it, published so you do not have to do it yourself.

Download the data: CSV · JSON — 27 columns per state, plus sources, assumptions and both sensitivity tables in the JSON. Free to reuse with attribution and a link back. Please cite the version and build date, because these numbers move with electricity prices.

Where the sauna number comes from

The inputs

EIA labels the Table 5.6 values preliminary estimates based on a cutoff model sample, and we carry that caveat forward — these are EIA's preliminary numbers, not final ones. We use the January–May year-to-date average rather than May alone, because a figure meant to be annualised should not sit on one month. A five-month average still carries winter-weighted rates where prices are seasonal; the dataset is a partial-year average, not a calendar-year one, and says so.

The arithmetic

kWh per session = heater_kW × heat_up_hours
                + heater_kW × session_hours × maintenance_duty_cycle

kWh per year    = kWh per session × sessions_per_week × 52

cost            = kWh × (state price in cents/kWh ÷ 100)

Central: 9 × 0.75 + 9 × 0.75 × 0.30 = 8.775 kWh per session, × 2.5 × 52 = 1,140.8 kWh a year, × $0.1811 = $206.59.

A warning for anyone quoting that $123–$340 band. It varies three inputs, and only two of them are technical uncertainty — heat-up time and duty cycle. The third, sessions per week, is a usage choice. Hold sessions at the central 2.5 a week and the band attributable to technical uncertainty alone is $153.62 to $283.40. If what you want is "how uncertain is the running cost of a 9 kW sauna", that narrower band is the honest answer; the wider one is uncertainty plus lifestyle.

Why this reads higher than the per-session figures in our home sauna cost guide. The closest row in that guide is a large electric barrel at 5–8 kWh a session, $0.80–$1.40; it also lists a 2-person infrared cabin at $0.25–$0.51 and a 4–8 kW traditional heater at $0.50–$1.20. We model 8.78 kWh, above the top of the barrel row, because this heater is 9 kW and we charge a full 45 minutes of it at 100% draw. The guide counts heat-up too, so that is not the difference — the difference is heater size and how long we hold it at full draw. The rest is the tariff: at the guide's $0.17/kWh our session is $1.49 rather than $1.59. Which is why a single site-wide "cost per sauna session" number is not a thing we are willing to publish.

The two assumptions doing the work

The 30% duty cycle is not ours, and it is a design target rather than a measurement. SaunaTimes states it directly — an electric sauna "performs best when the stove is 'on' no more than 30% of the time", and in the comments, "we want our heater to cycle on/off about 30% of the time". That is guidance for a correctly-sized heater in an indoor room. Ours is an outdoor barrel fighting much larger heat losses, which the same author is explicit about qualitatively but does not quantify. So the high scenario lifts the duty cycle to 45% — and that 45% is our assumption, with no source behind it.

The heat-up range is our reading, not a published range. The three published 9 kW claims we could find say "Guaranteed to deliver 175F in under 45 minutes" (vendor marketing), "we can get up to serving temp within 45 mins. to an hour", and "60-80 minutes for it to come up to temp" for a high-stone-mass heater. Those claims span 45 to 80 minutes. Our 30-minute low is an interpretation of "under 45", not something any source states. Our 60-minute high is a practical ceiling for a standard barrel setup, which means it understates the top end a reader of those claims would set — if you run a heavy-stone heater, your heat-up may exceed our high scenario.

Neither of these is hidden in a footnote in the data either. Both ship in the JSON assumptions block, in the same words.

Where the plunge number comes from

It has no electricity in it at all

A chiller-less plunge is a tub of water with ice in it. No pump, no compressor, nothing to plug in — so no state's electricity price touches it. The dataset carries plunge_electricity_kwh_per_year = 0 in all 51 rows, and that is zero by construction rather than an estimate that happened to land on zero.

Its running cost is ice, plus a little water. At the national average ice is just under 90% of it, and in the forty-one states where the model buys any ice at all the median is about 89%.

The physics has no unknowns

Cooling water with ice is a closed energy balance — nothing fitted, nothing estimated:

m_ice = m_water × cp × (T_tap − T_target) ÷ (Lf + cp × T_target)

with cp = 4.186 kJ/(kg·K), Lf = 334 kJ/kg, temperatures in Celsius, ice entering at 0 °C. Bagged ice out of a freezer is usually colder than 0 °C, which makes it slightly more effective per pound, so modeling it at exactly 0 °C mildly overstates how much you need. We would rather err in that direction and say so.

At EPA's national average inlet temperature of 61.5 °F, cooling 100 gallons to 50 °F takes 59.4 lb of ice — about three 20 lb bags — for $14.85 of ice plus $1.68 of water = $16.53 a fill. Fifty-two fills is $859.47 a year.

The two weak inputs, exposed rather than buried

Per-state tap water temperature is not published anywhere. EPA publishes one authoritative national figure — a weighted national average inlet temperature of 61.5 °F. To get a state signal we estimate:

state tap temp ≈ 61.5 °F + (state annual mean AIR temp
                            − population-weighted national annual mean AIR temp)

Air temperature is NOAA nClimDiv statewide means, 1991–2020; the national anchor is population-weighted with Census Vintage 2024 populations so it is consistent with EPA's own population-weighted figure. The assumption, plainly: that a 1 °F difference in annual mean air temperature means a 1 °F difference in annual mean mains temperature. Anchoring on EPA's national number cancels any constant offset between mains and air; what is left is a slope assumption of 1.0, and that is ours and is unverified. Real mains temperature is damped relative to air, and varies with source, pipe depth and season. NOAA publishes no DC record, so DC borrows Maryland's air temperature — its electricity price is its own real EIA figure.

So do not take our per-state estimate on trust. Look up your own tap temperature and read the ice cost straight off this table:

Add $1.68 a fill for water at EPA's national combined rate of $16.83 per 1,000 gallons ($7.74 water + $9.09 wastewater, 2024 estimate). Everything scales linearly with volume — multiply by (your gallons ÷ 100).

There is no published national price index for bagged ice. This is the least reliable input in the whole dataset. Retail listings we observed on 27 July 2026 ran from roughly $2.29 for a 20 lb bag at a warehouse club, as reported, up to $5.79 at home-improvement retail. We model at $5.00 a bag and publish the $3/$4/$5/$6 columns above precisely because a single figure here would be false precision. Ice price is a parameter, not a fact.

The zero-ice artifact — read this before quoting a cold state

In ten jurisdictions — AK, ID, ME, MN, MT, ND, NH, VT, WI, WY — the estimated annual mean tap temperature is already at or below the 50 °F target, so the model returns zero ice and shows water only: $87.52 a year.

An ice bath is not free in Alaska. It means the annual mean sits at or under the target, and an annual mean hides the season that matters. Summer tap water in every one of those ten states runs well above 50 °F and will need ice. Every affected row carries plunge_annual_mean_at_or_below_target = True and a note saying so — if you quote one of those states, quote the flag with it.

We publish no summer figure. Doing it properly needs seasonal mains temperature, and mains water is strongly damped seasonally, so applying our 1.0 slope to summer air temperature would materially overstate it. That would be a worse assumption than the one we have already made, so we left it out rather than manufacture it.

What this changes about the sauna-vs-plunge decision

Most "sauna vs cold plunge" content has the running-cost comparison backwards, because the plunge has no plug and therefore looks free. It isn't. Bagged ice at a few dollars a bag, three bags a fill, once a week, is a bigger annual line than heating a 9 kW sauna twice a week in most of the country.

Three things fall out of the table that are worth more than the headline ratio:

The sauna's cost is a map of electricity prices, nothing else. Every state burns the same 1,141 kWh a year. North Dakota to Hawaii is a 3.7× spread and all of it is the tariff. If you are deciding between sauna types, your state's ¢/kWh matters more than the brand.

The plunge's cost is a map of climate. Warm-state readers pay for the same 50 °F twice over — they start further away from it. Florida's estimated 76.9 °F tap water needs 138.9 lb of ice a fill against 59.4 lb at the national average, which is why Florida's plunge-to-sauna ratio is 10.76× while New York's, on 7.7 lb a fill, is 0.56×. The national 4.16× is a real number about the average household and a bad number about any particular one.

This is the strongest argument for a chiller that exists — a chiller replaces the ice line entirely, and the ice line is the whole bill. We are not putting a figure on that comparison. We do not own a chiller and have no sourced consumption data for one, and inventing a plausible-looking kWh figure would defeat the point of the exercise. If you are shopping that way, our cold plunge chiller roundup covers the hardware on specs and owner reports, without a running-cost claim we cannot support.

What this dataset does not do

• No chiller model. Stated three times because it is the most likely misreading.
• No seasonal breakdown. Everything is an annual mean, and both models swing seasonally — the sauna's duty cycle rises in winter, the plunge's ice need falls.
• No standby or idle draw. Heat-up plus session only.
• No purchase, installation, water treatment, filtration or maintenance cost. Running cost only. Purchase and install are covered in our home sauna cost guide and the DIY cold plunge setup walkthrough.
• No top-up ice or ambient reheating between fills. The plunge model charges ice once per fresh fill and nothing more. An outdoor tub warms back toward ambient within hours, so anyone plunging more than once on a fill, or re-icing a tub that has warmed up, spends more than this shows.
• No claim that the 30% duty cycle matches our specific outdoor barrel. It is a published design target applied here as an assumption.

The last two omissions both push the same way: for a household that plunges more often than once a week, or re-ices between sessions, the gap between plunge and sauna is wider than 4.16×, not narrower. We say so rather than wait for someone to find it. It does not follow that the gap is wider for everyone — sauna four times a week and plunge once and your own ratio drops below the national figure, which is exactly why the per-state, per-usage bases are printed rather than hidden.

Method, sources and citation

Every number here is either read out of a cited public dataset or is arithmetic on those figures using an assumption written down on this page.

To cite it: BankrollZen, US home sauna + cold plunge running-cost dataset, version 1.0, built 27 July 2026 — https://bankrollzen.com/blog/sauna-cold-plunge-running-cost-by-state/

Free to reuse with attribution and a link. If you spot an input you think is wrong, that is the point of publishing the working — the assumptions are all individually adjustable and the two weakest ship with sensitivity tables so you never have to take our central value on trust.

More on the gear itself: home saunas · cold plunge · contrast therapy protocols

Last reviewed: August 2026