The main types of maintenance in manufacturing are reactive, corrective, preventive, usage-based, condition-based, predictive, prescriptive, reliability-centered maintenance (RCM), and total productive maintenance (TPM). They differ by what triggers the work: a breakdown, a known defect, a calendar date, a usage count, a measured condition, or a forecast. Most plants run a mix, matching each strategy to a machine's criticality and failure pattern instead of applying one approach to every asset.
Understanding the types of maintenance in manufacturing matters because the strategy you choose sets your downtime, your maintenance labor, and your parts spend. Run everything to failure and you pay for emergency repairs and lost production. Service everything on a fixed calendar and you replace good parts early while still missing machines that wear faster than the schedule assumes.
This guide is for maintenance managers, plant managers, and manufacturing engineers building or updating a maintenance program. It covers each maintenance type, what it requires, and how to match strategies to the equipment on your floor.
Each maintenance type is defined by its trigger. Here is a quick reference:
Every maintenance type falls into one of two groups: reactive work that happens after something breaks, and proactive work planned to stop it from breaking. Moving work from the first group to the second is the single biggest cost lever in most maintenance programs.
The savings are well documented. The U.S. Department of Energy's O&M Best Practices Guide estimates that a well-run preventive maintenance program saves 12% to 18% compared to reactive maintenance, and that a functioning predictive maintenance program saves another 8% to 12% over preventive maintenance alone.
That doesn't mean every machine needs the most advanced strategy. A floor fan can run to failure. A five-axis CNC running a bottleneck job cannot. The goal is the right strategy per asset, which the sections below break down.
Reactive maintenance, also called run-to-failure or breakdown maintenance, means fixing equipment only after it fails. It requires no planning and no upfront cost, which is why it remains common. The trade-off is unplanned downtime, overtime labor, rush-shipped parts, and secondary damage when one failed component takes out others.
Corrective maintenance fixes a problem that has been identified but hasn't caused a breakdown yet. A technician spots a hydraulic leak during a PM inspection, or a vibration alert flags a loose mount, and a repair gets scheduled. It is often confused with reactive maintenance, but the difference is timing: corrective work is planned around production instead of forced by a failure.
Preventive maintenance (PM) schedules inspections, lubrication, and part replacements at fixed calendar intervals, such as monthly or quarterly. It is simple to set up and far cheaper than running to failure. Its weakness is that it ignores how the machine was actually used. Two identical mills on the same 90-day schedule get serviced at the same time even if one ran three shifts and the other ran one.
For a deeper comparison, see our guide on preventive vs. predictive maintenance.
Usage-based maintenance, also called meter-based maintenance, schedules service after a set number of cycles, strokes, or run hours. It fixes the core flaw of calendar-based PM by tying maintenance to actual machine use. A press gets serviced at 250,000 strokes; a CNC spindle gets inspected at 2,000 run hours.
Manual counter readings are the weak point. When usage data is captured automatically by a machine monitoring system, usage-based PM becomes reliable. Our guide on moving from calendar-based to usage-based maintenance walks through the switch, and the same data can help predict tool breakage.
Condition-based maintenance (CBM) triggers work when a measured value shows that equipment health is declining. Sensors track vibration, temperature, current draw, pressure, or fluid condition, and maintenance happens when a reading crosses a set threshold. Work only happens when the machine needs it.
Learn more in our condition monitoring guide.
Predictive maintenance (PdM) uses condition data trends and failure history to forecast when a component will fail, so work can be scheduled just before it does. Where condition-based maintenance reacts to a threshold, predictive maintenance looks at the trend line and estimates remaining useful life.
Prescriptive maintenance goes one step past predictive. It forecasts a failure and recommends what to do about it, such as reducing spindle speed to extend bearing life until a planned shutdown. It relies on extensive sensor data and advanced analytics, so it is mostly found in large, data-mature operations today.
Reliability-centered maintenance (RCM) is a framework for deciding which maintenance type each asset should get. Teams analyze each asset's functions, failure modes, and the consequences of failure, then assign reactive, preventive, condition-based, or predictive maintenance accordingly. RCM takes significant upfront analysis but produces a program where effort matches risk.
Total productive maintenance (TPM) is a lean manufacturing approach that makes equipment care a shared responsibility between operators and maintenance. Operators handle daily tasks like cleaning, lubrication checks, and visual inspections, a practice called autonomous maintenance. Maintenance technicians focus on planned and technical work. TPM measures success with OEE (overall equipment effectiveness).
Choose a maintenance type for each machine based on two factors: what a failure costs and how the machine fails. Critical machines with usage-driven wear belong on usage-based PM. Critical machines that show warning signs belong on condition-based or predictive maintenance. Low-cost assets can run to failure.
Most plants end up with a layered program: TPM habits on the floor, usage-based PM on production machines, condition monitoring on rotating equipment, and reactive maintenance for everything that isn't worth scheduling.
Each step up from reactive maintenance requires better data. This is where many maintenance programs stall: the strategy is chosen, but the data to support it isn't there.
Machine monitoring closes most of these gaps. Caddis captures cycle counts, run hours, downtime, and condition data such as vibration, temperature, and amperage directly from each machine. Its built-in preventive maintenance tool uses that data to trigger usage-based and condition-based maintenance automatically, and the same data feeds OEE tracking for TPM programs.
Move up one level at a time, starting with the machines that cost the most when they fail. Trying to jump from reactive maintenance straight to predictive usually stalls because the data and habits aren't in place yet.
For more on building a program around real machine data, see The Machine Maintenance Guide and our overview of what a CMMS is.
The four most commonly cited types are reactive, preventive, predictive, and condition-based maintenance. Broader lists also include corrective, usage-based, prescriptive, reliability-centered maintenance (RCM), and total productive maintenance (TPM). Most manufacturing plants use several of these at once.
Reactive maintenance happens after equipment breaks down. Corrective maintenance fixes a defect that was found before it caused a failure, often during an inspection or through monitoring, so the repair can be scheduled around production.
Preventive maintenance is scheduled on a fixed interval, either a calendar date or a usage count. Predictive maintenance uses condition data and trends to forecast when a component will fail, so work happens just before it's needed. Predictive requires more data but avoids servicing parts that still have life left.
It depends on the asset. For critical production machines, usage-based, condition-based, and predictive maintenance usually deliver the lowest total cost because they prevent unplanned downtime without over-servicing. For cheap, non-critical equipment, reactive maintenance is often the most cost-effective choice.
Usage-based maintenance schedules service after a set number of cycles, strokes, or run hours instead of a calendar date. It ties maintenance to how hard a machine actually works, which makes it a strong fit for CNCs, presses, and injection molders. It works best when usage data is captured automatically from the machine.
The types of maintenance in manufacturing range from reactive repairs to predictive and prescriptive strategies, and the right choice depends on each machine's criticality and how it fails. Most plants get the biggest gains by moving critical production machines off calendar schedules and onto usage-based and condition-based maintenance driven by real machine data. Book a demo to see how Caddis triggers maintenance from live machine data.
