What size battery keeps a refrigerator running?

The refrigerator is the appliance most people want backed up first, and it is also the one where the published specs mislead you in a specific, fixable way. Three different wattage numbers appear on a refrigerator spec sheet, only one of them belongs in a battery sizing calculation, and the largest of the three is the one people reach for. This guide does the arithmetic from published figures and shows the working. We have not tested any of these products, and the Amazon links above are affiliate links, disclosed at the top of this page.

Three numbers, one of them right

Take GE's GFE26JYMFS, a 25.7 cubic foot French-door model. Its spec page publishes all three:

• Volts, hertz, amps: 120V, 60Hz, 15A. This is the branch circuit requirement. Multiplying 15 A by 120 V gives 1,800 W, which is roughly six times what the appliance actually draws. This number is for the electrician, not for you.
• Watts: 313. This is the appliance's draw while the compressor is running.
• Energy consumption: 700 kWh per year. This is the EnergyGuide figure, and it is the one to size capacity against, because it already contains the compressor's duty cycle.

The conversion is one division. A year is 8,760 hours, so 700 kWh per year is 700,000 Wh divided by 8,760, or about 80 W on average. That 80 W already accounts for the compressor being off most of the time. A refrigerator does not draw 313 W continuously; it draws roughly 313 W for part of each cycle and close to nothing between cycles, and the annual figure averages that out for you.

Here is the same arithmetic across models whose annual kWh their makers publish:

| Model | Type | Published kWh per year | Average watts (kWh / 8,760) |

|---|---|---|---|

| GE GTE22JTNRWW | Top freezer, 21.9 cu ft | 405 | 46 W |

| LG LTCS24223S | Top mount, 23.8 cu ft | 438 | 50 W |

| Samsung RF30BB6600QLAA | French door, 30.1 cu ft | 695 | 79 W |

| GE GFE26JYMFS | French door, 25.7 cu ft | 700 | 80 W |

| GE GSS28NYYFS | Side by side, 27.9 cu ft | 732 | 84 W |

| Whirlpool WRS588FIH | Side by side, 28 cu ft | 747 | 85 W |

The pattern is worth pausing on: a large side-by-side with a through-the-door dispenser averages roughly twice a top-freezer of similar capacity. If you are choosing what to plug into a battery during an outage, the fridge you own already decided most of the answer.

What the kWh figure is, and is not

The label number is a test result under conditions that are not your kitchen. The DOE test procedure states its intent plainly: "to simulate typical room conditions (72 °F (22.2 °C)) with door openings, by testing at 90 °F (32.2 °C) without door openings." So the test runs hot and never opens the door, on the theory that the two effects trade off. The FTC's own consumer guidance is equally direct that the label is "only an estimate, based on typical use and a national average price for energy."

During an outage your kitchen may be hotter than 72 degrees F, and you will open the door. Both push consumption up. Treat the average-watts figure as a planning number with a stated basis, not a guarantee, and leave margin.

Runtime arithmetic

Runtime, at the ceiling, is rated capacity divided by average load. The table below does that division for five current power stations against four of the average-watt figures above. The assumptions are stated and they matter: these use each unit's rated Wh, not usable Wh, and they ignore inverter losses, the battery's low-voltage cutoff and any self-consumption by the station itself. Real runtime is lower. None of the makers publishes an inverter efficiency figure for this load, so we do not apply a derate rather than invent one.

| Unit | Rated capacity | At 46 W | At 50 W | At 80 W | At 85 W |

|---|---|---|---|---|---|

| DELTA 3 Plus, Elite 100 V2 | 1,024 Wh | 22.3 h | 20.5 h | 12.8 h | 12.0 h |

| Anker SOLIX C1000 | 1,056 Wh | 23.0 h | 21.1 h | 13.2 h | 12.4 h |

| BLUETTI Elite 200 V2 | 2,073.6 Wh | 45.1 h | 41.5 h | 25.9 h | 24.4 h |

| EcoFlow DELTA Pro 3 | 4,096 Wh | 89.0 h | 81.9 h | 51.2 h | 48.2 h |

Read those as ceilings and shade them down. The useful shape of the table is the ratio, not the decimal: a 1 kWh unit is a same-day answer for one refrigerator, a 2 kWh unit is a same-day answer for a fridge and a freezer, and covering a multi-day outage on one refrigerator alone takes 4 kWh or a way to recharge.

The startup surge, and the honest state of the evidence

Every compressor draws a brief spike when it starts, far above its running watts. That spike is what trips an undersized inverter, and it is the one specification the appliance industry does not publish. We checked consumer spec sheets, product pages and installation manuals from GE, LG, Samsung, Whirlpool and Frigidaire. None lists locked-rotor amps, starting amps or a surge figure. Every number circulating online traces to a compressor OEM datasheet, a trade forum or a reseller listing rather than to the company that sold you the fridge.

What is published comes from the other side of the plug, and it does not agree with itself:

• Generac publishes 700 running watts for a refrigerator or freezer and no starting figure at all, with a footnote that "starting watts are typically 2X running watts, but check your product literature for actual running and starting wattage requirements."
• Champion publishes a range: 150 to 400 W running against 800 to 1,200 W starting, and defines starting watts as "typically 1-2 times the running wattage" lasting "a fraction of a second." The chart and the definition imply different multipliers.
• EcoFlow and Jackery both publish 2 to 3 times running wattage on their own knowledge pages.
• Queensland's Electrical Safety Office, a state regulator, publishes "around 1,800 starting watts, 180 watts normal running" and says appliances with motors "can use up to ten times their listed wattage when starting up."

That spread runs from 2x to 10x. No US federal source publishes a refrigerator inrush figure. The disagreement is the finding, and the practical response is headroom rather than precision: against a few hundred running watts, a station rated 1,000 W or more continuous with a genuine surge rating above that is buying insurance against a number nobody published for your model. That is why the output column, not the capacity column, is where you should refuse to economize.

Boost modes do not count as surge headroom

The larger of the two wattage figures on a power station's box is often a software mode, and on two of these brands the maker states in its own documentation that the mode does not apply to a refrigerator.

BLUETTI's Elite 100 V2 user manual describes Power Lifting as follows: "This mode is only for pure resistive loads rated 1,800W-2,700W. Elite 100 V2 can handle such loads, its actual output power is still 1,800W." A refrigerator compressor is an inductive load, not a resistive one, so Power Lifting is not the number to size it against. BLUETTI publishes a separate 3,600 W surge figure for that unit, and that is the relevant one.

Anker's SurgePad carries a comparable caveat in its own FAQ: "SurgePad activates when total output exceeds rate output. Heat-generating devices work best. SurgePad does not support precision instruments and other devices with voltage protection or a strict voltage requirement." Anker's 2,400 W is that mode, and Anker publishes no separate hardware surge rating for the C1000, so its 1,800 W continuous is the number to plan against.

EcoFlow's X-Boost works the same way and is published separately from the surge figure: the DELTA 3 Plus lists 1,800 W total with a 3,600 W surge, and 2,200 W as maximum device power with X-Boost. The surge number and the boost number are different claims.

The four hours you may not need to cover

Before buying anything, check the arithmetic against how long your power actually goes out. USDA guidance published on FoodSafety.gov states that a refrigerator "will keep food safe for up to 4 hours during a power outage" with the door kept closed, and that "a full freezer will hold a safe temperature for approximately 48 hours (24 hours if it is half full and the door remains closed)." It also instructs that after an outage you should "never taste food to determine its safety."

If your utility restores power inside four hours, the published guidance is that a closed refrigerator handles it and a full freezer handles far more. The case for a battery starts where those windows end: multi-day storm outages, medically important refrigerated items, a chest freezer full of food, or an area where restoration is measured in days rather than hours.

Matching a unit to the job

• One refrigerator, a storm-length outage: the 1 kWh class. The EcoFlow DELTA 3 Plus, BLUETTI Elite 100 V2 and Anker SOLIX C1000 sit within 32 Wh of each other in capacity and all publish 1,800 W continuous, which clears a few hundred running watts with wide margin.
• You want the surge figure to be a real one: the DELTA 3 Plus and Elite 100 V2 each publish a 3,600 W surge alongside 1,800 W continuous, a 2x hardware ratio. Anker's headline second number is a boost mode instead.
• A refrigerator and a chest freezer: roughly double the capacity, which is where the BLUETTI Elite 200 V2 lands at 2,073.6 Wh, with 6,000+ cycles to 80 percent, the longest cycle rating of these five.
• Multi-day, or a fridge plus other circuits: the EcoFlow DELTA Pro 3 at 4,096 Wh, which is closer to installed-system scale and is covered alongside four other units in our plug-in backup guide.
• You expect to do this often: look at cycle life rather than capacity, and read LFP vs NMC chemistry. All five units here use LFP cells.

If the real question underneath this one is whether to solve it with a plug-in box or a wired system, portable power station vs home battery is the direct comparison, and how long can a home battery power a house runs the same arithmetic at whole-house scale. For why rated capacity and usable capacity differ, see depth of discharge.

Every product figure above is a manufacturer datasheet claim recorded with its published basis. The runtime numbers are arithmetic on those claims under the assumptions stated, not measurements we made, and not a promise about your refrigerator.