Coin cell mAh is not runtime: why same-size cells differ
The mAh number on the packet is a test result, not a promise. Four things between that number and the date you next open the battery door.
No, a bigger mAh number does not guarantee longer life. Capacity is measured at one specific low drain, at room temperature, down to one specific cutoff voltage. Change any of those three and the same cell gives a different runtime. That is why one CR2032 lasts a decade in a motherboard and a day in an LED tealight.
Does a higher mAh rating mean the battery lasts longer?
Only when both cells face the same load. A CR2032 is rated 220 to 240 mAh and a CR2025 160 to 170, so on paper the thicker cell should run about 40 percent longer in the same holder. In a remote that sips current, it does. The rating stops being a guide the moment your device stops behaving like the test rig.
The cutoff voltage is the part nobody prints on the blister pack. The rating counts charge delivered down to a threshold the test chose, not the threshold your gadget quits at. A circuit that gives up at 2.5 volts throws away charge that a circuit happy down to 2.0 volts would still be spending. Same cell, same rating, two very different runtimes.
What actually decides how long a coin cell lasts?
| Same-size pair | Rated capacity | What actually sets the runtime |
|---|---|---|
| CR2032 vs CR2025 | 220-240 mAh vs 160-170 mAh | Whether the holder is molded to 3.2 mm. A CR2025 sits loose in a CR2032 slot and loses contact, which ends the runtime long before the charge runs out. |
| CR2032 vs BR2032 | 220-240 mAh vs 190 mAh | Temperature and pulse current. BR chemistry runs roughly -30 to +80 C on a lower plateau near 2.8 V under load; CR has the stronger pulse for radios. |
| CR2450 vs CR2477 | 550-620 mAh vs 1000 mAh | Height and contact. The 5.0 mm CR2450 usually cannot reach both contacts in a 7.7 mm holder, and where it does, expect about 60 percent of the runtime. |
| CR2430 vs CR2450 | 270-300 mAh vs 550-620 mAh | Duty cycle. A transmit-only fob gets two to four years from a CR2430; a proximity key that listens constantly gets one to two. |
| One CR2032, four jobs | 220-240 mAh in every case | 5 to 10 years on CMOS backup, 2 to 4 in a car fob, about 1 in an AirTag, 24 to 48 hours in an LED tealight. |
Why does the same CR2032 last ten years in one device and a day in another?
That last row is the whole argument. One cell type, one capacity rating, runtimes that differ by more than a thousand to one. Nothing about the battery changed. The current draw did.
The CR2477 shows the same arithmetic from the other end. At typical beacon drains of 10 to 20 microamps average, its 1000 mAh runs five years or more, and at a 15 microamp average it works out to over seven. The microamps set the years. The milliamp-hours only scale them. Before you compare two capacity numbers, work out which side of that ratio your device sits on.
What a pulse load sees instead of capacity
A key fob transmitting, an AirTag chirping or a bike sensor waking up does not draw a steady trickle. It asks for a brief burst. Every cell has internal resistance, so a burst pulls the terminal voltage down for as long as it lasts. If that dip drops below the device's cutoff, the fob fails to transmit or the sensor resets, on a cell that is still most of the way full. Cold makes it worse, because internal resistance climbs as temperature falls. A remote that fails in a January car park often works again indoors.
This is why our BR2032 notes call it usable in a key fob but not ideal. Same 20 by 3.2 mm case, same 3 volts, 190 mAh against the CR2032's 220 to 240, and a carbon monofluoride chemistry that delivers pulses less strongly on a plateau near 2.8 volts under load. In a tire sensor that bakes and freezes all year, the BR is the correct cell and the lower number is irrelevant. In a remote you press from across the room, the CR wins. CR2032 vs BR2032 has the rest.
Why you cannot tell how full a lithium coin cell is
Lithium coin cells hold a flat plateau for most of their life, then fall away fast. A healthy CR2032 reads 3.2 to 3.3 volts open circuit and keeps reading close to that for years, so a meter tells you almost nothing about how much is left. Two readings do mean something: anything under 3.0 volts straight out of the packet was dead stock, and a device that misbehaves under load has already hit the cliff.
Shelf time works the same quiet way. A lithium coin self-discharges only about 1 percent a year, and a genuine Panasonic, Murata, Renata, Duracell or Energizer cell keeps over 90 percent of its charge for a decade. No-name multipacks routinely test 20 to 40 percent down on arrival. For a cell you fit into a thermostat and forget about for five years, that gap decides the outcome, not the number on the packet.
So how should you choose a replacement?
Match the code on the old cell first, because fit and chemistry outrank capacity every time. Where the holder genuinely has the room, a taller cell is a real gain: CR2450 at 550 to 620 mAh against CR2430 at 270 to 300 in the same 24.5 mm diameter, or CR2477 at 1000 mAh where 7.7 mm of depth exists. Where it does not, a thin cell shimmed into a deep holder is a contact failure waiting to happen, and CR2032 vs CR2025 covers that pair in full.
Buy from a brand with a readable date code, and treat unbranded capacity claims as marketing. Chemistry choice in 1.5 volt watch cells is a separate question with its own answer in why watch batteries die fast. The full size chart lists every cell we track.