
When an entrant collapses inside a tank, the next few minutes depend on choices made weeks earlier: who comes, how fast, and whether they have practiced in a space like this one. A confined space rescue plan is where those choices are written down.
OSHA's permit space rule spells out what that plan must settle. A phone number on the permit does not.
Table of Contents
In general industry, the rule is 29 CFR 1910.146, Permit-Required Confined Spaces. Paragraph (d)(9) makes rescue part of the written permit space program. The employer must develop and implement procedures:
Paragraph (k) covers choosing a service (k)(1), training in-house rescuers (k)(2) and retrieval systems (k)(3). Three scope limits come first:
Appendix F to 1910.146, OSHA's non-mandatory guide to evaluating rescuers, is direct: "Merely posting the service's number or planning to rely on the 911 emergency phone number to obtain these services at the time of a permit space emergency would not comply with paragraph (k)(1) of the standard."
An outside service can still qualify: Appendix F says the employer must at a minimum contact it to plan and coordinate the evaluations.
A 2012 study in the Journal of Occupational and Environmental Hygiene shows why, combining federal fatality data, a survey of 21 large companies and senior fire officers' estimates. "Twelve (57%) of 21 surveyed companies reported that they relied on the fire department" for permit space emergency response (Wilson, Madison and Healy, 2012).
Arrival was quick: "Median fire department arrival times were about 5 min for engines and 7 min for technical rescue units." The full rescue was not.
Fire officers estimated rescue times of 48 to 123 minutes, or 70 to 173 with hazardous materials. The authors concluded that "fire departments usually are not able to effect a confined space rescue in a timely manner" in a life-threatening emergency. They proposed that fire departments support "a properly trained and equipped on-site rescue team."
Paragraph (k)(1) sets five duties for every employer that designates rescue services under (d)(9):
Timely depends on the space. Appendix F counts response time from notification to being set up and ready for entry. For an IDLH atmosphere, or a space that can quickly develop one, the team or service would need to stand by at the space. Where the danger is limited to mechanical hazards causing injuries such as broken bones, "a response time of 10 or 15 minutes might be adequate."
Appendix F also has employers measure the team's performance "during an actual or practice rescue." A critique follows.
Both are allowed.
| Question | In-house rescue team | Outside rescue service |
|---|---|---|
| Who trains the rescuers | The employer, under (k)(2) | The service; Appendix F asks whether it meets (k)(2) |
| Response time | Can stand by at the space | Depends on distance; Appendix F asks what happens if it becomes unavailable mid-entry |
| Knowledge of your spaces | Built through practice in your spaces | Needs hazard briefings and site access under (k)(1)(iv)-(v) |
When the employer's own employees are designated rescuers, (k)(2) requires the employer to:
Representative spaces must simulate the opening size, configuration and accessibility of the real ones. Appendix F reads the annual practice as applying "provided that the team or service has not successfully performed a permit space rescue within that time."
Paragraph (k)(3) requires retrieval systems or methods whenever an authorized entrant enters a permit space. The exception: "unless the retrieval equipment would increase the overall risk of entry or would not contribute to the rescue of the entrant."
Under (k)(4), if an injured entrant was exposed to a substance with an MSDS or similar written information required at the worksite, it goes to the treating medical facility.
OSHA does not publish a rescue plan form. This template is the article's own summary of (d)(9), (k) and Appendix F. Fill in one copy per permit space, or per group of spaces with the same hazards and configuration.
| Plan element | What to record for this space | Basis |
|---|---|---|
| Hazards and response time | The space's hazards, the response time they allow, and whether standby is needed | (k)(1)(i), Appendix F |
| Rescue team or service | Who is designated, when they were evaluated, and how they are summoned | (d)(9), (k)(1) |
| Hazard briefing and access | When the team was told the hazards and given access to the space | (k)(1)(iv)-(v) |
| Training records | PPE proficiency, rescue duties, entrant proficiency, first aid and CPR | (k)(2)(i)-(iii) |
| Practice | Date and type of the last practice rescue and its critique | (k)(2)(iv), Appendix F |
| Retrieval system | Harness, line, anchor point or device, or the reason retrieval is not used | (k)(3) |
| Medical information | Which MSDS or written information goes with an injured entrant | (k)(4) |
| Unauthorized rescue | How the crew prevents would-be rescuers from entering | (d)(9) |
Review it when a space, its hazards or the rescue arrangement changes.
An annual practice tests the team once. The knowledge behind it has to hold the rest of the year: each space's hazards, the rescue plan, the retrieval rules and the entrant-level knowledge (k)(2)(ii) requires.
Blank Slate is a cognitive-readiness platform built on AI-optimized spaced retrieval. Spaced retrieval combines spaced repetition (reviewing material over time) with retrieval practice (answering from memory instead of rereading). It works on top of the training content you already have.
Rescuers answer questions in short sessions (about 60 seconds), 3-4 times a week, adding up to a few minutes a week. Each answer feeds True Knowledge Level™, Blank Slate's real-time readiness metric, which shows what a person or team knows now.
Blank Slate has no published confined space study. The closest evidence is a study of US Air Force pararescue trainees presented at I/ITSEC 2023.
Blank Slate users' accuracy on the course knowledge in the app rose from 67% to 81% over 16 weeks, on about 100 seconds of use per week. App users also had an 84% first-time pass rate on an applied knowledge test, versus 46% for non-users (I/ITSEC 2023). That is an association, not proof of cause. The USAF Pararescue case study has the details.
Blank Slate does not deliver the training 1910.146 requires. It does not replace rescue training, first aid and CPR certification, PPE proficiency or practice rescues. It coexists with your LMS, which records who was trained (see the difference). The industrial safety page covers EHS teams.
An employer that has never asked its service how long a rescue in a specific tank would take has not met (k)(1), whatever the permit says.
For each permit space, either the evaluated service meets the response time its hazards allow, or the employer needs a team at the space. That team's readiness between practices is the employer's to show, and a Knowledge Gap Audit shows where it stands today.
No. OSHA's Appendix F to 1910.146 says merely posting a service's number or planning to rely on 911 would not comply with (k)(1). An outside service is allowed once the employer evaluates it and meets the rest of that paragraph.
At least once every 12 months, under 1910.146(k)(2)(iv), using simulated rescues in actual or representative spaces. OSHA's non-mandatory Appendix F reads the requirement as not applying when the team has successfully performed a permit space rescue within that time.
Every in-house rescuer must be trained in basic first aid and CPR under 1910.146(k)(2)(iii). Certification is narrower: at least one member of the rescue team or service holding current certification must be available.
Only when, under 1910.146(k)(3), the equipment would increase the overall risk of entry or would not contribute to the rescue. Otherwise every authorized entrant uses one, and vertical spaces over 5 feet deep need a mechanical device available.
A rescue plan is only as ready as the people named in it. A Knowledge Gap Audit shows what your team knows today, where gaps may exist, and where to focus first.