What wears out and when
What you have

What wears out in a house, and what a service life figure is worth

A published service life is a trade association's compiled estimate, not a prediction for your house. Here is how to use one anyway, and what your own evidence beats it with.

Start here Essentials

At minimum, put a published figure and a publisher beside each row you already have. The comparison of warranty periods against trade figures, the climate effects and the sinking fund material are Go deeper.

  1. Look up each row of your systems sheet in the service life table and write the figure and the publisher in the row (what a figure is).
  2. Subtract the install date from the figure and write the year money is likely needed. Use the reserve calculator if you want the arithmetic done (planning horizon).
  3. Mark every component whose age you could not establish as unknown, and give it the earliest plausible year with the reason beside it (unknown ages).
  4. Circle anything with three years or less remaining and carry it to the ranked job list.

Minimum time: about 40 minutes. Done when: every systems row has either a replacement year or the word unknown.

1. What a published service life figure actually is Understand

Somewhere in most people's heads is a sentence like "a water heater lasts about twelve years". It is a useful sentence. It is also worth knowing exactly what kind of sentence it is before you plan a decade around it.

The figures circulating in American home maintenance advice come overwhelmingly from compiled tables published by trade associations. The one this manual reproduces is InterNACHI's estimated life expectancy chart, from the International Association of Certified Home Inspectors, which gives expected service lives component by component for most of what is in a house.

Convention A figure like that is a trade association's compiled estimate. It is not a warranty, not a standard, and not a prediction about one house. It is closer to a typical value drawn from a large population, and it is being applied to a sample of one.

Limits

InterNACHI's chart carries no publication date and no methodology statement. Nothing on it says when it was compiled, from what data, or how the figures were derived. That is a real limit on how much weight any figure from it should carry, and this manual states it everywhere the chart is used, including on every row of the service life table. Use it as a planning starting point, name the publisher when you write a figure down, and let your own evidence overrule it whenever you have some.

Two sentences the manual will not write, and the honest versions beside them:

The tempting sentenceWhat the evidence supports
Your water heater will last twelve years.InterNACHI's chart estimates a service life for that type. Your unit's install date is on the plate. Those two numbers give a planning horizon, not a forecast.
The roof has three years left.The published estimate minus the age gives three. What the roof actually has left is a question for somebody who looks at it, and the published figure is the reason to have them look.

That second line is the whole use of a service life figure. It is not an answer. It is a trigger for a question. A component inside three years of its published life is a component to get an opinion on, to price, and to start funding. It is not a component to replace because a table said so.

2. What makes a real house diverge from the table

Four things move a real component away from a population average, and all four are things you can observe.

  • Water. Anything wet lasts less long than the same thing dry. A water heater in a damp basement, a furnace with a condensate problem, wood in contact with soil, a deck that never dries out. The drainage work in Keeping water out is also life-extension work for everything downstream of it.
  • Sun and weather. Roof coverings, exterior paint, sealants, decking and anything with a rubber or plastic component age by exposure as much as by time. A south-facing slope and a north-facing slope on the same roof do not age at the same rate.
  • How hard it runs. Equipment sized small for a house runs more hours. A heat pump that both heats and cools runs year round where an air conditioner runs five months. Hours, not years, are what wears a motor.
  • Whether anybody serviced it. This is the one you control. ENERGY STAR's maintenance checklist links a dirty filter directly to dust building up in the system, which it connects to expensive maintenance and early system failure. That is a published statement that maintenance affects equipment life, and it is as specific as the published sources get.

Physical principle None of those four changes the published figure. They change where your component sits relative to it, and the direction is usually knowable even when the amount is not. A furnace with seventeen years and no service records is not in the same position as a furnace with seventeen years and a sticker for every one of them, and you do not need a number to act on that difference.

One more divergence worth naming, because it shapes expectations more than any single component: the age of the housing stock itself. Harvard's Joint Center for Housing Studies reported a median US home age of 44 years in 2023, and that in that year average per owner maintenance spending on homes built before 1980 ran 76 percent higher than on homes built since 2010. That is a statement about what owners spent, not about what a house requires. What it is good for is calibrating surprise: on an older house, more rows and more spending is the pattern rather than the exception.

3. The evidence that beats the table

Three kinds of evidence outrank a published figure, and you collected all three in Walk your house.

  1. The age on the data plate or the installer's sticker. An observed install date turns an estimate about a population into arithmetic about your unit. It is the single most valuable thing on your systems sheet.
  2. Service history. Stickers, invoices, and what a technician wrote. A component that has been serviced annually for ten years has a different expected future from an identical one that has not, and you have the record to say so.
  3. Your own log. What you observed and when. A water heater you noted as "slight rust at base, photographed" in 2025 and "rust spreading" in 2026 is telling you something no table knows.

This is the argument for the upkeep log as a sheet rather than as a nice idea. A dated observation is evidence. A memory of having noticed something is not, and six months later it is not even reliably a memory.

Test it Where you have no install date and no service history, you can still produce evidence. Photograph the component today, from the same position, and write the date. In two years the pair of photographs is a measurement of a rate of change, which is exactly what you lack.

4. Turning a service life into a planning horizon

The arithmetic is deliberately simple, because the inputs do not deserve anything more elaborate.

StepWhat you writeExample
Published figure and publisherBoth, in the same row"12, InterNACHI"
Install dateObserved, or the word unknown"2016, data plate"
Years remainingFigure minus age, floored at zero"2 years, on that figure"
Year money is likely neededThis year plus the remaining years"about 2028"
Annual set-asideReplacement cost divided by the published lifethe reserve calculator

Every one of those entries carries the uncertainty of the one before it. A replacement year derived from an assumed install date is an assumption with arithmetic performed on it, which is why the row says where the date came from. What upkeep costs takes the annual set-aside column and sums it, which is the one place in this manual where a column of estimates turns into a single number, and it says so there too.

What the horizon buys you is sequencing. Three components with remaining lives of 2, 7 and 11 years are a plan: one to price now, one to fund, one to note. The same three with no dates are an anxiety.

5. Gradual failure and sudden failure

Two components with the same remaining life can need completely different treatment, because of how they fail.

  • Gradual. A roof covering degrades visibly over years. Exterior paint fails in patches. A deck board softens. You get warning, the warning is visible from the ground or from a walk, and the cost of being a year late is usually small.
  • Sudden, and wet. A water heater tank splits. A washing machine supply hose bursts. A sump pump fails during the storm it exists for. There is no warning, and the cost of the failure is not the component: it is the floor, the drywall, the contents and the drying.

That distinction is why The priority rule treats a tier 3 item whose failure causes a tier 1 problem as tier 1. A water heater in a finished basement is not an equipment item. It is a water item wearing an equipment label.

The useful consequence is that some very cheap parts protect very expensive outcomes:

The cheap partWhat it is protecting againstWho fits it
Braided stainless washing machine supply hosesA burst hose running until somebody comes homeMany householders; a plumber if the valves will not close
A drip pan with a drain under a water heaterA split tank reaching the floor and the contentsA plumber, usually at the next replacement
A fixture stop valve that actually turnsHaving to shut the whole house to deal with one sinkA plumber
A tested sump pump, and a note of which breaker it is onFinding out it does not work during the stormYou, by pouring water into the pit

None of those is a maintenance task in the usual sense. They are changes to how a failure lands, and they belong on your list as rows of their own.

6. When you cannot establish an age

This is the common case, especially on a house you did not buy new, and it has a procedure rather than a guess.

  1. Write unknown in the install column. Not an estimate. The word.
  2. Find a ceiling. The component cannot be older than the house, and often cannot be older than a renovation you can date. A furnace in a 1996 house is at most 30 years old.
  3. Find a floor if you can. A permit record, a brand that did not exist before a certain year, a fitting style, a previous owner's recollection recorded as hearsay.
  4. Plan on the earliest plausible year and write why. "Unknown; assumed at end of published life because no service history and visible corrosion" is a planning decision anybody can audit, including you in three years.
  5. Put a row on the job list for establishing it. A permit search, an invoice in the box of papers, or a technician's opinion at the next service visit. Many unknowns become known for free.

The reason to do it this way rather than writing "about 2010" is that the two entries behave differently over time. "Unknown, assumed at end of life" stays visibly an assumption. "About 2010" is a date within six months, and every calculation downstream of it looks solid.

Go deeper

Why a trade figure and a warranty period differ

A manufacturer's warranty period is a commercial commitment: it is the length of time the manufacturer will stand behind the product, priced accordingly. A trade association's service life figure is an estimate of typical useful life. They answer different questions and they are usually different numbers, with the warranty shorter. Neither is a prediction for your unit. When they disagree, the warranty tells you what the maker will pay for and the trade figure tells you roughly when to start saving.

Finish service lives, and where this manual stops

Floors, counters, cabinets, paint and tile have published service lives too, and they are mostly tier 5 in The priority rule: they look bad long before they fail at a job. Learn Home Finishes, another manual on this shelf, works through those materials and the same question about what a published figure is worth. This manual stays with the systems and the shell.

What happens to the figures in a hard climate

A hot humid climate puts more hours on cooling equipment and more stress on anything organic. A very cold climate puts more hours on heating equipment and cycles everything through freeze and thaw. A marine climate puts salt on metal. The published tables do not segment by climate, so the honest use of them in a hard climate is as an optimistic case: when your evidence and the table disagree, your evidence wins, and in a hard climate expect it to disagree downward.

A sinking fund is not an emergency fund

The reserve this manual builds is a sinking fund: money set aside for a known future expense on a known rough schedule. An emergency fund covers the unknown. They do different jobs and in practice people spend the second one on the first, which is how a planned roof becomes a crisis. What upkeep costs builds the sinking fund; the emergency fund is outside this manual, and nothing here is financial advice.

7. See it See it

Fictional example

Ray and Diane Okonkwo, the 1958 ranch outside Dayton. Their systems sheet after an hour with the service life table, in 2026:

SystemInstalledPublished figureRemainingRow reads
Gas furnace2009, installer stickerInterNACHInear or past itReplace by: soon; get an opinion at the annual service
Central air conditioner2009, same stickerInterNACHInear or past itReplace by: soon; price as a pair with the furnace
Gas tank water heater2016, data plateInterNACHIa few yearsReplace by: about 2028. In a finished basement, so tier 1
Asphalt shingle roofunknownInterNACHIunknownUnknown; previous owner said approximately 2011. Permit search on the job list

Known: three install dates, each from a label or a sticker Ray photographed during the walk.

Assumed, and the assumption is InterNACHI's rather than Ray's: that the published figures describe equipment like theirs. Their furnace is a 2009 unit in a basement that is dry, serviced irregularly, and running a mild-winter load; nothing in the chart knows any of that.

Still open, and who settles each: the roof's age (the city building department's permit history, which Ray has now requested); the furnace's actual condition at seventeen years (a heating technician's inspection, which is on the calendar anyway); and whether the water heater's position over a finished floor warrants a pan, which is a plumber's answer and a cost they have not priced.

Their conclusion, stated with its uncertainty intact: under these assumptions the furnace and air conditioner are the next large spend, potentially within two or three years, and the thing that would settle it is the technician's opinion at the annual service rather than any further arithmetic. The water heater is the one that gets moved up a tier, not because it is older, but because of where it sits.

8. Practice Practice

Minimum: three rows, looked up and dated, which is the Start here result. Time: about 40 minutes for three rows, 90 minutes for a full sheet. Materials: your systems sheet with install dates in it, and the service life table open.

  1. Work down the sheet one row at a time, writing the figure and the publisher together. A figure without a publisher is unusable in two years.
  2. Do the subtraction and write the year, not the number of years. "About 2028" ages correctly; "2 years left" does not.
  3. Mark each row gradual or sudden in the notes, and flag any sudden one that sits above or beside something expensive.
  4. Write a reason beside every unknown, and add a row to the ranked job list for establishing it.
  5. Photograph the two oldest components, dated, and put the photographs somewhere you will find them.
Go deeper

Two exercises for a house you intend to keep. Build the full replacement timeline: put every row's replacement year on one list in order, and look at what lands in the same two or three years, because that is the information a reserve exists to act on. And price the two nearest rows for real, with a local quote rather than a published figure, which turns the reserve calculator from an exercise into a number you can act on.

9. Your house Your house

10. Check your work Check your work

Criteria

  • Does every row with a figure also name the publisher of that figure?
  • Does every row without an install date say unknown, with a reason and a plan for establishing it?
  • Can you point to at least one row where your own evidence disagrees with the published figure, and say which you are planning on and why?
  • Have you marked which components fail suddenly and wet, and moved any of those that sit over something expensive?
  • Are your replacement entries years rather than counts of years?

Worked example

The Okonkwos pass on four criteria. Their rows name InterNACHI, their roof row says unknown with hearsay recorded and a permit search queued, they can point to the water heater as the row where position overrules age, and their entries are years. Where they are thin is the third criterion: they have no row where their own evidence contradicts the table, because they have no service history at all. That is itself a finding, and the honest entry in their notes is "no service records for any system; the published figures are the only evidence we have", which is a sentence that tells a future reader exactly how much to trust the sheet.

Common mistakes

  • Treating the figure as a date. A published life is a population estimate being applied to one unit. It triggers an opinion; it does not supply one.
  • Dropping the publisher. A bare number on a sheet is a number nobody can check, including you.
  • Replacing something because the table says so. A working component at the end of its published life is a component to price, fund and watch, not one to remove.
  • Ignoring the sudden-and-wet distinction. It is the difference between a planned expense and a claim.
  • Letting an estimate become a date. "About 2010" written once is 2010 forever.

What is still unresolved

A table cannot tell you how long your water heater has left, and this chapter does not pretend otherwise. What settles condition is somebody who looks: a heating technician at the annual service in Heating, cooling and hot water, a roofer, a plumber, or a home inspector. What settles cost is a local quote, which What upkeep costs needs and which this manual ships no built-in figures for. And what settles the order is The priority rule, where a component's remaining life is one input among several rather than the whole ranking.

11. Knowledge check and scenario

What does it mean that InterNACHI's chart carries no publication date and no methodology?
That you cannot tell when the figures were compiled, from what data, or how. It does not make them useless; it makes them a starting point that should travel with the publisher's name attached, and that your own evidence outranks whenever you have any.
Your furnace is two years past its published life and working fine. What does this chapter tell you to do?
Price it, fund it, and get an opinion at the next annual service. The published figure is a trigger for a question, not an instruction to replace. What it also tells you is to stop being surprised by the possibility, which is what a planning horizon is for.
Why does a water heater in a finished basement get treated differently from an identical one in a garage?
Because it fails suddenly and wet, and what it would damage is different. The priority rule moves a tier 3 item whose failure causes a tier 1 problem up to tier 1. The component's remaining life is unchanged; the consequence of the failure is not.
(Go deeper) Why is a manufacturer's warranty period usually shorter than a trade association's service life figure?
They answer different questions. The warranty is a commercial commitment the maker has priced; the service life figure is an estimate of typical useful life. Neither predicts your unit. The warranty tells you what the maker will pay for; the trade figure tells you roughly when to start saving.
You buy a house with no records at all. The water heater's data plate is legible and reads 2011, and there is a small rust bloom at the base. It is in a finished basement. What goes on the sheet?
This is most of the right answer and it leaves out the two things that matter. The rust is evidence that outranks the table, and the finished basement changes what the failure costs. A note is not a plan.
The photograph turns an impression into a measurement you can repeat next year. The tier 1 flag reflects where it sits rather than how old it is. And the plumber's opinion is the only thing that can tell you whether the rust is a fitting weeping or the tank itself, which is the whole question.
Possibly the right outcome, reached without evidence. Rust at the base can be a weeping fitting above it, which is a small repair, or the tank, which is not. Find out which before buying a water heater, and get the drip pan priced in the same conversation.

Checklist

Records your learning only; it is not a qualification.