Gas limit is actually two related settings on Ethereum and other EVM-compatible networks: the maximum computational effort a user authorizes for a single transaction, and the maximum total computation a block can hold. Both use "gas," the unit that measures the work needed to run code on the network, whether that is a plain transfer or a complex smart contract call.
At the transaction level, the gas limit is a safety ceiling, not a fee by itself. Wallets usually estimate it automatically: a simple ETH transfer needs 21,000 gas, while calling a decentralized application can need anywhere from tens of thousands to several hundred thousand units depending on the contract's logic. The actual cost equals gas used multiplied by the gas price, so a higher limit costs nothing extra since unused gas is never spent. The real risk is setting it too low: execution that runs out of gas fails and reverts, yet the network keeps the fee for work already performed, a state known as an "out of gas" error.
At the network level, validators collectively set a block gas limit that caps total computation per block, shaping throughput and the hardware nodes need to keep up. This ceiling has risen steadily through client-signaled community votes, reaching 60 million gas per block after Ethereum's Fusaka upgrade activated in December 2025, with further increases toward 100 million and beyond actively debated as part of ongoing scaling plans.
Because these decisions trade off decentralization against capacity, gas limit changes remain one of the most closely watched governance topics in Ethereum, alongside layer-2 networks that keep everyday transaction fees low regardless of what the mainnet limit allows.