If you've stopped using a credit card and want to keep it from being closed, the cadence that works everywhere is one posted charge every 90 days. That clears every major US issuer's inactivity threshold with a buffer. Stretch longer at your own risk — the bare-minimum windows differ by issuer, and the failure mode is permanent closure.
Once every 90 days, on every unused card. Any posted charge counts — $1 or $1,000, the issuer's review only checks the date. Quarterly clears Citi and BoA (6-month thresholds), Chase, Amex, Discover (12), and Capital One (24) with room to spare.
Skip the cadence math → or jump to pricingThe bare-minimum cadence by issuer
Inactivity thresholds vary. The rough boundaries:
- Citi, Bank of America — 6–12 months. Aggressive. Plan on quarterly.
- Chase, American Express, Discover — 12–18 months. Moderate. Every 6 months works but isn't generous.
- Capital One — 18–24 months. The most lenient. Once a year is usually enough.
If your portfolio spans issuers, your cadence has to match the most aggressive one — there's no point matching each card to its bank. One rhythm across the whole set is easier to maintain, and the rhythm that works for everyone is quarterly.
Why frequency matters more than amount
The issuer's inactivity-review system asks one question: when did this account last have a posted transaction? It returns a date. The date is compared against the threshold. Amount, merchant category, total spend — none of that enters the query.
So a $1 charge in month 11 saves a card whose threshold is 12 months. The same card with no activity gets closed. Spending $5,000 on a card in March and nothing after is functionally identical, for inactivity purposes, to spending $1 in March. Both produce the same "last activity date." Issuers don't reward heavy use with extra runway — they just track the most recent posting.
Why quarterly is the safe cadence
Even at Citi's aggressive 6-month threshold, a 90-day cadence keeps you at half the limit at worst. You're never closer to the wire than the buffer itself.
Quarterly is also the easiest cadence to remember. Four times a year, once per season. Pick fixed dates — January 1, April 1, July 1, October 1 — put them in a calendar, done. Vague reminders ("use this card sometime soon") fail. Specific dates work.
Why monthly is even safer
Some people prefer monthly use because monthly is easier to set-and-forget than quarterly. The easiest implementation: a small recurring subscription on each unused card with autopay. The subscription bills monthly, the charge posts, autopay clears the balance, the card stays active forever.
Good anchor subscriptions: Apple Music or Spotify ($10–15), iCloud+ or Google One ($1–10), NYT or WSJ ($4–25), Notion or 1Password ($3–10). Use what you'd be paying for anyway — the card becomes the rail, the subscription is the anchor.
The failure mode is the anchor breaking. Across multiple cards, the chance that at least one subscription gets canceled or repriced within a year is high. That's the case for delegating the cadence entirely.
Why you shouldn't push the yearly limit
Capital One often leaves cards open with one charge per year. Some Chase products tolerate 14–18 months. Pushing to the edge is technically possible — and a bad idea.
Yearly cadence requires perfect execution. Miss once and the card is closed. Issuers can also tighten internal timelines without telling you. The downside of being late is losing the card permanently; the downside of being early is nothing. The asymmetry says: be early.
What doesn't count as "using" the card
Three common misconceptions:
- Logging into the app. Doesn't count. Issuers track posted transactions, not engagement.
- Paying the balance to $0. Doesn't count. Payments are inbound credits, not posted charges. A $0-balance card looks more inactive to the issuer, not less.
- Receiving a refund. Doesn't count. A refund is a credit, not a charge.
What does count: a fresh outbound posting from a merchant. New purchase, new subscription bill, anything that adds a debit to the account.
Bottom line
Once every 90 days works everywhere. Once every 30 days is easier to maintain because monthly subscriptions are reliable. Once every 12 months technically works for some issuers but fails the moment you forget once.
Pick monthly if you want set-and-forget. Pick quarterly if you want minimum overhead and you're disciplined. Don't pick yearly — the savings aren't real and the failure cost is permanent.