Yes, steel corrodes. All common steels rust when exposed to oxygen and moisture, because iron thermodynamically prefers to exist as an oxide. Carbon steel can begin rusting within days in humid air, loses roughly 0.1 mm (about 4 mils) of thickness per year in a moderate industrial atmosphere, and can perforate in months in marine splash zones. Stainless steels corrode too, just far more slowly and only when their protective chromium oxide film breaks down. What changes across steel types is not whether corrosion happens, but how fast and by which mechanism.

The electrochemistry: why iron gives up electrons

Corrosion is a battery that short-circuits through the metal itself. At an anodic site on the steel surface, iron atoms surrender electrons and dissolve as ferrous ions (Fe²⁺). Those electrons travel through the steel to a cathodic site, where dissolved oxygen and water consume them. The two reactions only run if the electrons have a path through the metal and the ions have a path through an electrolyte — typically a thin film of water.

Rust is not a single compound but a family of hydrated iron oxides. Freshly formed ferrihydrite matures into lepidocrocite (γ-FeOOH), then converts over months to goethite (α-FeOOH); magnetite (Fe₃O₄) forms in low-oxygen conditions. Rust occupies roughly two to six times the volume of the iron it consumed, which is why corroding rebar cracks concrete from the inside. That expansion pressure is a structural failure mechanism in its own right, independent of metal loss. In plain language: rust is bulkier than the steel it eats, so it pries apart whatever confines it.

Three conditions must coincide for atmospheric corrosion: an anodic reaction, a cathodic reaction, and an electrolyte bridging them. Remove any one and corrosion stops. That single rule explains every mitigation strategy in this article. It also explains why dry steel in a desert or sealed steel in dry nitrogen survives for decades while identical steel fails in a coastal warehouse.

Measuring the cost: microns per year and what moves the rate

Engineers quantify corrosion as penetration rate, usually in micrometers or mils per year (1 mil = 25.4 µm), or as mass loss per area per time. The rates span nearly four orders of magnitude across environments and alloys:

Environment Carbon steel rate (typical) Stainless steel 304 rate Primary driver
Indoor, heated, dry Under 1 µm/yr Negligible Low humidity, no electrolyte film
Rural atmosphere 2–10 µm/yr Under 0.5 µm/yr Rain, oxygen, few pollutants
Urban or industrial 25–60 µm/yr 1–3 µm/yr Sulfur dioxide, particulates
Marine coastal 50–200 µm/yr 2–10 µm/yr Chlorides
Marine splash zone 200–500 µm/yr Pitting possible Wet-dry cycling, high chloride
Buried soil 10–50 µm/yr Pitting risk Soil resistivity, microbes
Seawater immersion 100–200 µm/yr Pitting risk Chloride, biofilms

These figures are conservative design ranges from corrosion-engineering references, not guarantees; real rates can swing several-fold with alloy grade, surface finish, and local conditions. Featured values in this table sit mid-range for temperate climates. Multiply the rate by service life to get expected metal loss. A 60 µm/yr industrial rate over a 25-year design life means 1.5 mm of steel gone — enough to halve a 3 mm structural section, so design codes add a corrosion allowance that thick.

The relative humidity threshold is the single most useful number. Atmospheric corrosion of steel is negligible below about 60% relative humidity (RH) because the surface cannot sustain a continuous water film. Above roughly 70–80% RH, rates climb steeply. Most of the corrosion in a typical building or vehicle therefore happens during a small fraction of hours, not evenly over time. The physical reason: below the threshold, adsorbed water exists only as isolated molecules; above it, water condenses into a connected film that can carry ions between anodes and cathodes. That connectivity, not the absolute amount of water, switches corrosion on.

The rates above are averages over surface area. Localized corrosion does not average: a single pit can penetrate millimeters per year while the surrounding surface shows almost nothing. This is why pool depth is the wrong mental model — think of stress concentration in a small area instead.

Where the simple story breaks: chlorides, galvanic couples, and oxygen cells

Chloride is the most aggressive common ion because it breaks passive films at specific, predictable thresholds. Stainless steel 304 typically resists pitting in chloride concentrations below roughly 200–500 parts per million; 316, with molybdenum, tolerates roughly 1,000–2,000 ppm. Seawater is about 19,000 ppm chloride, which is why 304 fixtures near the coast develop rust blooms on fasteners and welds. There is a caveat worth flagging: pit resistance also depends on temperature and pH, and chloride concentration at the bottom of a pit can concentrate many times above ambient as the pit geometry traps ions.

A 200 ppm chloride threshold means tap water in coastal regions and desalinated supplies can pit 304 when chlorides are above that line. Deicing salt, at millions of tons spread each winter in North America, explains why northern vehicle frames fail from inside out below where road spray collects. Concentration cells form under gasket edges, deposits, and ship's bilge areas, where oxygen access differs across the surface and drives local anodic attack. Stray direct-current leakage from electrical faults or welding equipment dissolves steel anodically at rates governed by current, not by chemistry. Rebar inside cracked concrete with chloride ingress pitting risk jumps sharply once the chloride level at the bar surface reaches roughly 0.4% by weight of cement for conventional reinforcement.

Galvanic corrosion is the second exception. When two metals contact in an electrolyte, the more anodic one corrodes faster than it would alone while the more cathodic one is partly protected. Steel coupled to copper, brass, or graphite suffers accelerated attack; the area ratio governs severity, so a small steel fastener in a large copper plate corrodes especially fast. There is a counterintuitive detail with real design consequences here: stainless steel, the "corrosion-proof" choice, is cathodic to carbon steel, so attaching 304 fasteners to a carbon-steel structure concentrates attack on the structure. The fastener survives; the beam doesn't. If the geometry were reversed — carbon-steel fastener in stainless structure — the fastener corrodes even faster, because the small anode carries the current from the large cathode.

Microbially influenced corrosion is the third exception, and the one that surprises inspectors most. Sulfate-reducing bacteria in anaerobic soil, tank bottoms, and water mains generate hydrogen sulfide that corrodes steel even where oxygen is absent. Rates in soils with active bacterial populations can reach 50–100 µm/yr. Handled correctly, steel buried in dry, well-aerated gravel behaves differently from identical steel in wet clay crawling with bacteria. The professional caveat: corrosion in anaerobic environments often proceeds via direct microbial-electron uptake without the usual oxygen cathodic reaction. Standard inhibition logic — cut the oxygen, stay safe — therefore fails, which is why cathodic protection on buried pipelines must account for microbial activity, not just moisture.

Common misconceptions, corrected with numbers

Four beliefs cause most of the wasted steel and unnecessary maintenance in service:

Copper does not "rust." Copper weathers into green patina through other chemistry; mistaking green stains for rust sends operators chasing the wrong metal. Steel rust stains are orange-brown; patina stains are blue-green.

Stainless does not mean stain-free. Type 304, the workhorse of appliances and railings, holds a passive chromium oxide layer only about 1 to 3 nanometers thick — tens of atoms. That film spontaneously reforms in seconds when scratched, as long as oxygen and enough chromium are present. But it cannot reform in crevices where oxygen is depleted, which is why chlorinated pools attack pool-rail welds and why passivation and proper weld cleaning are real specifications rather than marketing.

Rust does not self-arrest. The hydrous oxides form a porous, poorly adherent layer that lets water and oxygen reach the metal below. Unlike the aluminum oxide film, rust lacks the cohesive, dense structure that protects the underlying metal. Plain steel in wet conditions keeps rusting until mechanically consumed or until something stops the cycle.

Paint and coating failure starts at edges, welds, and beneath the film. Recoat cycles are determined by inspection of those hot spots, not by surface area, which is why spotting corrosion under coating matters more than topcoat condition.

Recognizing, measuring, and interrupting the process

The practical section is a decision framework for anyone responsible for steel assets. Each step names a threshold or measurement, not a vague instruction.

Step one: identify the exposure class. Dry interior steel below 60% RH needs little action beyond quality paint. Humid interiors, coastal zones, road-salt exposures, and immersion zones each demand a different strategy. Classify the environment honestly before selecting any mitigation; misclassification wastes money in both directions.

Step two: inspect for the signature of localized attack. Pitting shows as small dark spots or rust blooms on machined or stainless surfaces; uniform attack shows as a dull, thinned face on carbon steel. Carriage bolts, weld toes, crevices, and thread roots are the inspection zones with the highest failure statistics, because they trap water most effectively.

Step three: measure where it matters. Ultrasonic thickness gauging tells you metal loss; visual assessment alone routinely underestimates internal wall loss, particularly in low-point sections of piping and tank bottoms. Schedule inspections at intervals consistent with observed rates — more often where rates exceed 50 µm/yr.

Step four: select the countermeasure by mechanism. Coatings physically block the electrolyte. Galvanizing (zinc at 30–85 µm typical coating thickness on structural sections) works both as a barrier — zinc oxide films the surface — and as a sacrificial anode that protects exposed steel at interruptions, because zinc gives up electrons more readily than iron. Cathodic protection supplies electrons externally; it is the method of choice for buried pipelines and tank internals and works even where coatings fail. Alloy selection (316 or duplex grades) buys chloride tolerance at higher material cost. Design changes — draining low points, eliminating crevices, keeping dissimilar metals apart or isolating them with insulating gaskets — often cost the least and last the longest. Choose the countermeasure that matches the actual mechanism, or you will pay for the wrong protection.

Step five: establish a maintenance trigger. Recoat before visible red rust develops, when spot coating loss or undercutting is first detected. Replace structural members at calculated section-loss thresholds, not at failure. Cathodic protection requires periodic potential surveys to remain effective. The galvanic-anode versions must be sized in mass to provide the required current for the design life; undersized anodes run out and silently leave the structure unprotected.

The cost frame: restoring steel structures runs into hundreds of millions of dollars annually in developed economies, with individual large bridge recoating projects in the tens of millions of dollars. A cleaning, priming, and painting cycle for a typical industrial structure costs hundreds of dollars per square meter of surface. That is why preservation decisions are economic as well as mechanical, and why records of rate and exposure class conversion are worth maintaining.

The limits of what can be known

Corrosion prediction carries irreducible uncertainty, and the numerical ranges in this article are design guidance, not predictions for a specific asset. Reported rates from different sites and references vary by factors of two to five for nominally identical steel because surface condition, trace contaminants, local weather, and inspection methods differ. Where the evidence needed to pin down a rate or a threshold is absent, the honest professional default is conservative design plus periodic inspection rather than a false-precision estimate.

Severity also changes across a service life. Rates that average 20 µm/yr over a decade can spike seasonally if localized conditions change, such as when a drain clogs or a coating develops a pinhole. Chloride thresholds in particular are not sharp universal constants; they depend on grade, pH, oxygen availability, and temperature, so published values should be treated as bands within which behavior transitions, not as hard floors.

The final constraint is money. Corrosion decisions are resource allocations, and the correct question is rarely whether steel will corrode but whether the protection purchased is proportionate to the consequence of metal loss at that specific location. Aesthetic rust on a garden gate and wall loss on a bridge member carry different costs, so the same mechanism and the same scientific table lead to different practical answers.