Safeguard
Compliance•

How do you prove the annual backup restore test under NYDFS?

NYDFS 500.16(d) requires testing your ability to restore critical data from backups at least annually. What to restore, measure and record as proof.

Safeguard Security Team
3 min read

Section 500.16(d) requires covered entities to test, at least annually, their ability to restore critical data and information systems from backups, in addition to testing incident response and continuity plans. Prove it with a dated record of what you restored, how long it took, who did it, and what failed.

This is general information, not legal advice.

What the text says

Three connected duties sit in section 500.16:

  • Maintain backups necessary to restore material operations, protected from unauthorized alteration or destruction.
  • Test the incident response and business continuity plans at least annually with the staff and management critical to the response.
  • Test the ability to restore critical data and information systems from backups, also at least annually.

The business continuity plan must also include procedures for backing up information essential to operations with sufficient frequency and storing it offsite.

Design a test that means something

A restore of one small file proves little. A better design:

  1. Choose a system that matters, such as a customer facing database.
  2. Restore it to an isolated environment, not over production.
  3. Start from the backup copy that would be used in a real event, including one held offsite.
  4. Verify integrity: checksums, row counts, an application level smoke test.
  5. Time each stage and compare against the recovery time objective in your asset inventory.

If your objectives were never measured, this is the first time you learn the real number.

What to record

ItemWhy it matters
Date and participantsShows the annual test happened and who took part
Systems and backup sets usedShows the scope was critical data
Elapsed timesCompares reality with objectives
Verification method and resultShows data was usable, not just present
Failures and fixesShows learning and follow through
Approval of resultsLinks to CISO oversight

Protect backups from the same attacker

Ransomware plans often fail because backups were reachable from the compromised environment. Section 500.16(e) asks that backups be adequately protected from unauthorized alterations or destruction. Separate credentials, immutability features where available, and offsite copies are the usual answers. Test that the backup credentials are not the same ones an attacker could take from production.

Mistakes to avoid

Backup programs fail quietly. Jobs report success while the data is unusable. Restores work for the database but not for the secrets and configuration needed to run it. The restore is tested once, then the architecture changes. Encryption keys are stored only in the system being restored. Verify integrity after each test, include configuration and keys in the scope, repeat the test after major architecture changes, and keep a copy of keys somewhere that survives the incident.

Quick check

Can a tabletop exercise replace the restore test?

No. The rule lists testing of incident response and continuity plans and, separately, testing the ability to restore critical data and systems from backups.

Do I need to restore everything every year?

The text says critical data and information systems. Scope is a risk decision that you should document.

Sources

Last reviewed: 19 October 2026.

Never miss an update

Weekly insights on software supply chain security, delivered to your inbox.

Self-healing security runs on Safeguard.

Your first fix PR is minutes away.

No sales call required, even your agent can complete the purchase over MCP.