Pagefile on SSD Causes Wear: The Math That Settles It

Where the myth comes from

Early SSDs deserved the caution. First-generation consumer drives from 2008–2012 shipped with small NAND endurance budgets and immature wear-leveling firmware, so every avoidable write genuinely counted, and moving the pagefile to a spinning disk was a reasonable hedge. The advice then froze in time and kept circulating long after the hardware changed underneath it.

Modern consumer drives are a different species. A typical 500 GB TLC drive carries a warranty endurance of 150–300 TBW — terabytes that may be written over its life — and the usual replacement cycle for a system SSD is measured in years of exactly that kind of abuse. The number that matters is printed on the spec sheet; folklore is not on the spec sheet.

The arithmetic nobody runs

Run the numbers the myth never does. Pagefile traffic on a normal desktop with adequate RAM is single-digit gigabytes per day, and even a heavy case — a memory-hungry workload on 8 GB — sits at 10–20 GB of pagefile writes daily. Call it 20 GB a day, every day, all year: that is roughly 7 TB of writes annually.

• Rated endurance of a typical 500 GB consumer SSD: 150–300 TBW under warranty
• Realistic pagefile write traffic: 1–10 GB per day; 20 GB/day is the heavy case
• 20 GB per day, all year ≈ 7 TB of writes annually — the heavy case
• 150 TBW ÷ 7 TB/year ≈ 20+ years of the heavy case before the warranty endurance is spent
• Real drives routinely outlive their TBW rating: the warranty figure is a floor, not a death date
• Ground truth on your machine: the SMART attribute Percentage Used in CrystalDiskInfo or the vendor's health tool

What disabling the pagefile actually costs

The commit limit of Windows equals installed RAM plus pagefile size, and applications reserve commit — they do not politely ask what is free. With no pagefile, a game or an Adobe project that reserves 25 GB against a 16 GB machine fails allocation with an out-of-memory error while the real RAM sits half empty. Crash analysis dies too: with no pagefile.sys there is no memory dump after a blue screen, so the next BSOD is a mystery by design.

And the gain is zero, which is what makes this trade so lopsided. Moving the pagefile to a mechanical drive to "save" the SSD trades microsecond SSD latency for millisecond HDD latency on every page-out — a measurable slowdown purchased to avoid wear the arithmetic above already dismissed. Leave the pagefile on the SSD, leave the size system-managed, and spend the worry on what actually hurts the drive: heat and a full disk.

Questions and Answers

Read next