CapEx and maintenance reserves: how much to actually budget for a rental (2026)
Jun 11, 2026 · 10 min read
CapEx is where marginal deals go to die. A property can "cash flow" $150 a month for three years and then hand the entire gain back in one afternoon when the HVAC fails. Sellers omit the line entirely; most investors copy a 5% default that has nothing to do with the age of the roof. Here's the component math that produces a number you can defend — and what that number does to NOI, DSCR, and the deal.
Maintenance vs CapEx: two budgets, two tax treatments
Maintenance(repairs) keeps the property in its current condition: the $180 service call, the $90 garbage disposal, the drywall patch after a tenant moves out. It's frequent, individually small, and — usefully — deductible in the year you pay it.
Capital expendituresreplace or improve whole components: a roof, a furnace, a kitchen. They're rare, individually large, and depreciated over 27.5 years rather than deducted immediately (the de minimis safe harbor lets you expense items up to $2,500 per invoice, which catches appliances and water heaters for most small landlords). The distinction matters at tax time — the full breakdown is in rental property tax deductions — but it matters even more in underwriting, because the two lines behave differently. Maintenance is a smooth, predictable drip. CapEx is a cliff. Budgeting them as one blended "repairs" percentage is how investors end up surprised by an expense they could have scheduled five years in advance.
Why percent-of-rent rules quietly fail
The common defaults — 5% of rent for maintenance, 5-10% for capex — share one fatal assumption: that wear scales with rent. It doesn't. A 30-year architectural shingle roof on a 1,400 sq ft house costs about $11,000 to replace whether that house rents for $1,100 in Cleveland or $2,800 in Phoenix. At 8% of rent, the Cleveland house banks $1,056/year toward capex; the Phoenix house banks $2,688 — for the same roof, the same furnace, the same water heater on the same clock.
The result is systematic: percentage rules understate capex on cheap properties and overstate it on expensive ones. And since low-price, high-yield properties are exactly where spreadsheet cash flow looks best, the error concentrates in the deals that can least afford it. This is the same family of mistake as treating the 50% rule as an underwrite instead of a triage tool — fine for a first pass, dangerous as a final answer.
The component method: derive the reserve, don't guess it
For each big-ticket component, divide replacement cost by useful life. Here's the full schedule for a typical 1,400 sq ft, 3-bed single-family at 2026 contractor prices:
- Roof (architectural shingle): $11,000 ÷ 25 years = $440/year
- HVAC (furnace + condenser): $9,000 ÷ 18 years = $500/year
- Water heater: $1,800 ÷ 10 years = $180/year
- Kitchen (cabinets, counters, appliances): $14,000 ÷ 20 years = $700/year
- Bathrooms (2 × $7,000): $14,000 ÷ 20 years = $700/year
- Flooring (LVP throughout): $7,000 ÷ 12 years ≈ $583/year
- Interior paint (full repaint): $3,500 ÷ 6 years ≈ $583/year
- Windows: $9,000 ÷ 30 years = $300/year
- Exterior (siding, gutters, driveway allowance): $8,000 ÷ 25 years = $320/year
- Electrical / plumbing allowance: $6,000 ÷ 30 years = $200/year
Total: about $4,500/year, or ~$375/month— before a dollar of routine maintenance. On a $1,600/month rent, that's 23% of gross income for capex alone, which is why the 5-10% defaults feel comfortable and underwrite wrong. Add $80-150/month for routine maintenance (more for older systems and rougher tenant classes) and the honest combined line on this archetype runs $450-500/month full-cycle.
Two fair adjustments before you panic. First, full-cycle assumes you own through every replacement; if you buy a house with a 5-year-old roof and sell in year ten, the roof never hits your ledger — though it hits your sale price instead, because the buyer's inspector runs the same math. Second, a recently renovated property genuinely earns a lower near-term reserve — which is the legitimate version of the argument every listing agent makes. The way to capture both: walk the property, estimate each component's remaining life, and re-run the division. The rehab cost estimator prices the component replacements, and the framework in how to estimate rehab costs covers the walkthrough itself.
What honest reserves do to a real deal
A $220,000 single-family renting for $1,950/month. 25% down, $165,000 loan at 7% over 30 years (P&I ≈ $1,098/month). Fixed expenses: $230/month taxes, $110 insurance, 7% vacancy ($136), 8% property management ($156).
- With the listing-flyer assumption ($100/month maintenance + capex): NOI ≈ $14,616/year, cap rate 6.6%, cash flow ≈ +$120/month, DSCR ≈ 1.11
- With component-derived reserves ($300/month on this property's actual ages): NOI ≈ $12,216/year, cap rate 5.6%, cash flow ≈ −$80/month, DSCR ≈ 0.93
One line item moved this deal from "modest cash flow" to "pays you nothing and doesn't cover its own debt on collected income." That's not an argument that the deal is bad — it's an argument that the reserve assumption is the underwrite on marginal deals, the same way the vacancy line is in what vacancy rate to assume. Run your own numbers both ways in the cash-on-cash calculator before trusting either one.
The NOI convention trap
Here's a wrinkle that bites investors comparing cap rates: by appraisal convention, NOI is calculated before capital expenditures. Brokers and seller pro formas follow that convention enthusiastically, because excluding capex inflates NOI and therefore the implied value at any given cap rate. Your own underwrite should reserve for capex anyway — the cash leaves your account regardless of where accountants file it — but when you compare your numbers to a listing's, make sure both sides use the same convention. A seller's "7% cap" with no capex line and your 7% cap with $300/month reserved describe two very different properties. The NOI calculator and cap rate calculator keep the definitions straight, and reading a pro forma covers the other six places seller math drifts optimistic.
How much cash to hold, and when to fund it
The monthly reserve answers "what does this property really earn?" A separate question is "how much cash do I need in the account?" — because capex doesn't arrive smoothly. Three layers:
- Liquidity floor: six months of PITIA. On the deal above, roughly $8,600. This is the buffer that turns a dead HVAC plus a vacant month from a crisis into a bad quarter. Many DSCR lenders independently require 3-6 months of verified reserves at closing.
- Age-weighted capex funding at purchase. For each component: replacement cost × (age ÷ lifespan). A 9-year-old water heater on a 10-year life means ~$1,620 of its $1,800 replacement should be banked on day one — the previous owner consumed that life, and the inspection is where you find out. Sum the shortfalls across components; on older properties this number routinely hits $8,000-15,000 and belongs in your cash-to-close math right next to closing costs.
- The monthly drip. Auto-transfer the component-derived amount to a separate account every month and treat it as spent. If it sits in the operating account, it reads as cash flow — and gets spent as cash flow.
Four mistakes that show up constantly
Double-counting with the 50% rule. The classic heuristic already includes maintenance, capex, and vacancy. Applying 50% and itemizing a reserve counts the roof twice and kills deals that pencil. Pick one structure per pass.
Letting the rehab subsidize the reserve — forever. A full gut renovation legitimately buys you low capex for 5-10 years. It does not buy you a $50/month capex line for a 30-year hold, because flooring, paint, and appliances cycle two or three times in that window even when they start new.
Scaling single-family numbers to multifamily by doormat count. A duplex shares one roof but carries two kitchens, two baths, and often two furnaces and water heaters. Per-unit capex on small multifamily runs 75-90% of a comparable single-family — not 50%. The shared-structure discount is real but smaller than it looks.
Confusing deferred maintenance with capex.The $12,000 of work the inspector finds is not a reserve item — it's purchase price. Negotiate it, fund it at closing, or walk. Reserves are for the components that are fine today and won't be in 2031.
FAQ
How much should I budget for capex on a rental property?
Derive it from component lifespans rather than picking a percentage. Sum each big-ticket item's replacement cost divided by its useful life — roof, HVAC, water heater, kitchen, baths, flooring, paint, windows, exterior — and you'll land around $330-400/month full-cycle on a typical 1,400 sq ft single-family, before routine maintenance. Most percent-of-rent defaults (5-10%) understate that badly on low-rent properties, because a roof costs the same whether the house rents for $1,100 or $2,800.
What is the difference between maintenance and capital expenditures?
Maintenance (repairs) keeps the property in its current condition — fixing a leak, patching drywall, servicing the furnace. CapEx replaces or improves whole components — a new roof, new HVAC, a kitchen renovation. The IRS treats them differently too: repairs are deductible in the year you pay them, while capital improvements are depreciated over 27.5 years (with a de minimis safe harbor that lets you expense items up to $2,500 per invoice). In your underwrite, budget both: roughly $80-150/month for routine maintenance plus a separate capex reserve.
Is the 1% rule a good way to estimate maintenance costs?
The old heuristic — budget 1% of property value per year for maintenance — is a tolerable sanity check on mid-priced properties but breaks at the extremes. On a $150K Midwest rental it suggests $1,500/year, which won't cover one water heater plus a service call in a bad year. On an $800K coastal property it suggests $8,000/year, which likely overshoots. Component math beats value math because components, not prices, are what wear out.
Does capex count against NOI?
By appraisal convention, no — NOI is calculated before capital expenditures, which is why listing pro formas love it. In your own underwrite, you should absolutely model a capex reserve as a recurring monthly cost, because the cash leaves your account either way. Just know which convention a number uses before you compare cap rates: a seller's 7% cap with zero capex and your 7% cap with $250/month reserved are not the same deal.
How big should my cash reserve be at closing?
A practical floor is six months of PITIA (principal, interest, taxes, insurance, association dues) plus an age-weighted capex fund: for each major component, multiply replacement cost by its age divided by its lifespan, and hold the shortfall. A 9-year-old water heater on a 10-year life means you should already have ~90% of its $1,800 replacement banked on day one. Many DSCR lenders also require 3-6 months of PITIA in verified reserves just to close.
Put the reserve in context
CapEx is one line, but it compounds through everything downstream — NOI, cap rate, cash flow, DSCR, and whether the deal verdict is real or borrowed from a future roof. The full TrueCap analyzer carries your maintenance and capex assumptions through the entire underwrite in one pass, so you can flip between the listing's number and the component math and watch the whole picture move. Related reading: how to underwrite a rental in 60 seconds, estimating rehab costs, and what vacancy rate to assume.