Methodology

How we research costs, permits and data

The short version

  • Our permit figures come from the open data Minneapolis and St. Paul publish. We request no owner names or addresses, and we publish only counts and summary figures, never the permit text.
  • A permit’s declared value is the figure a city uses to set the permit fee. It is not a contract price, so we never use it to judge a quote.
  • Rules sort each permit by the kind of bathroom work its description names. We tested how often they get it right, and we report the results with their limits.
  • The calculator combines 556 declared Minneapolis permit values and 92 published figures in Twin Cities 2026 dollars. It gives a project subtotal, selected upgrades and an optional reserve.

Where the permit figures come from

We pull residential building permits that mention a bathroom from the cities’ public open-data services: the City of Minneapolis CCS Permits dataset (data through October 9, 2026) and the City of Saint Paul Approved Building Permits dataset (which ends June 30, 2025). The pull used on this site was retrieved on October 13, 2026 and is kept unchanged, with a checksum for every file, so every figure can be rebuilt from the same raw records. We request only non-personal fields, never owner names, addresses, parcel numbers or contractor names. The permit descriptions are free text written by applicants and city staff; we publish only counts and quartiles, never the descriptions.

Each permit is counted once (by permit number). New-construction and addition permits are set aside. Then rules (permit-classifier-v2) read each description and give it one of eight categories: a full-room scope stated in the text; a bathroom remodel whose extent isn’t stated; tub or shower work only; single components or a half bath; a new bathroom in the house; a bathroom within a larger project; no bathroom work; or too little text to tell.

The rules match whole words in context, so “upstairs bathroom” is not stair work, “waterproofing” is not roofing and a grab bar is not a wet bar, and they skip the code notes and trade-permit lists that cities append to descriptions.

Our main group is bathroom remodels, full-room or unspecified scope, because the text usually can’t tell a gut from a refresh: Minneapolis cuts descriptions at 255 characters and many say only “remodel bathroom.” For each group we report the number of permits, how many carry a declared value above $0, and the 25th, 50th (median) and 75th percentiles of those values.

A declared valuation is not a contract price. Under Minnesota Rules 1300.0160 the applicant’s estimated job value, accepted by the building official, is used to calculate permit fees. These records contain no contract prices, so they can’t tell anyone how a declared value compares with what a homeowner paid; we don’t compare them, and we don’t use them as a check on quotes. Separate plumbing and electrical permits aren’t counted, many tub and shower swaps pull no building permit at all, a permit can cover more than one bathroom, and an issued permit doesn’t mean the work was done. See the permit data report for the tables.

How we checked the sorting rules

Rules that read free text make mistakes, so we measured them. We drew a stratified random sample of 200 permits, half from each city. In each city, 60 were ordinary descriptions and 40 were spread as evenly as the permits allowed across 11 trigger words (about 4 each) that misled our first version (waterproofing, upstairs, grab bars, conversions, kitchens, windows and similar). That over-samples ordinary descriptions and rare trigger words and under-samples common ones such as windows, kitchens and basements: a trigger word appears in 60% of Minneapolis and 56% of St. Paul permits that mention a bath, but in 40% of the sample.

The sample was split in advance: a development half, which the rules were written from, and a holdout half, which was kept aside to measure them once. Permit descriptions that our privacy screen (automated patterns plus an AI review) flagged as naming a person or showing an email address, phone number or street address were left out of the sample.

Two separate Claude AI instances labeled each sampled permit from a written guide, without the rules, our statistics or the other set of labels in view. No human review of those labels has been completed. The two sets of labels agreed on 194 of 200 permits (Cohen’s kappa 0.96). The AI agent that wrote the sorting rules settled disagreements in the development half with a written reason; the 4 disagreements in the holdout half were left out of the measurement rather than relabeled. Those are among the hardest permits to read, so leaving them out may make the rates below a little high.

On the 96 remaining holdout permits, 96% of the permits the rules put in the main group matched the agreed AI label (46 of 48; 95% interval 86–99%), and the rules found 100% of the permits assigned to that group by the agreed AI labels (92–100%). Across all eight categories, the rules matched the agreed AI labels for 93% (86–96%). Our first version’s “whole-room” group scored 87% precision and 74% recall on the same permits.

These figures measure agreement with AI-generated labels, not independently established real-world accuracy. Counting each holdout permit for the number of permits in its sampling group gives estimates for the roughly 4,582 existing-home permits that mention a bathroom and passed our privacy screen: the main group’s precision is about 96% (approximate 95% interval 79–99%, 48 permits; Minneapolis 94%, interval 70–99%; St. Paul 100%, interval 80–100%), and agreement across all eight categories is about 95% (interval 83–99%; Minneapolis 97%; St. Paul 90%, interval 75–96%). The permit report has the full table.

Limits of that measurement: the plain rates describe the sample, and the weighted ones are estimates whose intervals are wide and only approximate, because some sampling groups have one to three holdout permits; permits that our privacy screen dropped could not be sampled; and two AI label sets agreeing can reflect shared errors as well as consistency.

The AI agent that prepared the annotation guide had already read all 200 sampled descriptions, holdout included, for the privacy screen, so the holdout figures may be somewhat optimistic. The rules were not changed after the holdout was measured. Errors seen there (an unlisted room type such as an exercise room, a closet borrowed for space, a bathroom mentioned only as a boundary, a description that opens with asterisks) would be fixed in a later version, which would need a new holdout. The asterisk error also moves 19 published permits between categories; the permit report shows which.

How the calculator’s price ranges are built

We collected 803 figures from Minneapolis contractor permits and prices published by remodelers and cost guides. We removed duplicates, full remodels with values low enough to suggest cosmetic work, and unusually high or low amounts. That left 648 figures from 14 sources. Of those, 556 are declared Minneapolis permit values from 2021 to 2026, and 92 are published prices. Each published range counts once, at its midpoint.

To compare cities, we adjusted the labor share using local bathroom-trade wages from the U.S. Bureau of Labor Statistics and treated materials as nationally priced. To bring older figures to 2026 dollars, we used the bureau’s construction wage and residential construction materials indexes.

The typical price is the estimated median for the project, room group and finish. We estimate size and finish differences by comparing prices within the same source. The spread comes from variation across all 556 Minneapolis permit valuations after accounting for project differences. We apply that same spread to each typical price, about 74% to 135% of its value, and round to the nearest $500. This estimates the middle half using the pooled permit data. An individual contractor’s quote depends on the work and materials it includes.

A full remodel needs tub or shower work plus at least two other parts. Moving walls and moving plumbing together count as one part. Wall work needs a separate quote. We use the same starting subtotal for compact and standard rooms because many source records omitted the area and few identified a compact bathroom. Full remodels from 60 to 150 sq ft share the next subtotal. Evidence is thinner at the large end of that band. Above 150 sq ft, we give a quote checklist for a full remodel.

The tiled shower or tub replacement subtotal is the same for a given finish in every full bath. It covers a replacement shower, a tub replaced with another tub with tiled walls, or a tub converted to a tiled shower. A separate tub that stays in place does not change the price. Finish adjustments come from published cost guides with named tiers. A permit value alone cannot identify what materials a homeowner chose.

A half-bath remodel, panel shower system, walk-in tub, new bathroom or partial remodel outside these rules gets a quote checklist. If a tub or shower project is paired with a half bath, we ask you to select the room that has the tub or shower. The calculator’s version is est-2026-11-19.v6.

Upgrades and known conditions use separate local allowances. Some come from a higher-priced remodeler and may run high for a lower-priced contractor. The optional reserve adds 10% at the low end and 15% at the high end after selected allowances, rounded to $100. Quotes should list materials, glass, fees and exclusions for your home.

All 9 planning ranges

These are 2026 Twin Cities project subtotals before upgrades and reserve. Use the calculator to price your selections.

Full bathroom remodel

RoomBasicMidrangeHigh-end
Under 60 sq ft$16,500–$30,000$19,500–$35,500$36,500–$66,000
60–150 sq ft$24,000–$43,500$28,500–$52,000$52,500–$95,500

Project subtotal only. Upgrades, known conditions and the reserve are added in the calculator.

Tiled shower or tub replacement

RoomBasicMidrangeHigh-end
Any room with a tub or shower$7,500–$14,000$9,000–$16,500$16,500–$30,500

Project subtotal only. Upgrades, known conditions and the reserve are added in the calculator.

Download the nine planning ranges (CSV). Each row gives the low and high estimates in US dollars. The file includes the model version and a link back to the method.

Suggested citation: PermitPriced, “2026 Twin Cities bathroom remodel planning ranges,” model est-2026-11-19.v6. Link to the range tables.

Sources and how we check claims

  • We prefer primary sources: statutes and rules (revisor.mn.gov), city permit pages, the Minnesota Department of Labor and Industry, the EPA, Minnesota Housing, and companies’ own published prices. Every guide lists its sources with the date we accessed each one, and the costs page lists the sources behind every figure it shows.
  • Guides link to sources for building and electrical codes, safety rules, loan limits and legal rights. Verification of these consequential claims is still in progress; consult the linked authority or a qualified professional before relying on a rule or eligibility statement. Permit statistics are filled in from the dataset at build time, so the displayed figures stay consistent with that dataset.
  • A source citation is not a guarantee that a rule is current or that every statement has been independently checked.
  • A page’s “updated” date changes only when the page changes. A source being listed does not by itself mean every sentence on the page was verified against it.
  • Our guides never average prices from different sources into one “average cost”; they show each source’s range and explain the differences. The calculator’s ranges are the one place we combine sources, through the price model described above.
  • We don’t publish reviews, testimonials or rankings of remodelers.

Corrections

If you find an error, email testLG@gmail.com with the page and the source. We correct verified errors promptly and note substantive corrections on the page.