The complaint usually arrives in the same words: "It made me start again from the beginning." The learner did forty minutes, closed the window, came back the next day and landed on slide one. Sometimes the quiz answers vanish as well, and nothing looks wrong in the course itself, which makes it maddening to diagnose. The cause almost always involves SCORM suspend data, bookmarking or the way the session ended. Why won't a SCORM course resume where the learner left off, and how do you fix it for good?
A SCORM course fails to resume when its saved state is too large, never saved, or never read back. SCORM stores that state in a field called cmi.suspend_data, limited to 4,096 characters in SCORM 1.2 and 64,000 characters in SCORM 2004 3rd and 4th Editions. A course that exceeds the limit, exits without asking the LMS to suspend the attempt, or loses its final commit when the browser closes can restart from the beginning. The fix is usually to save less, commit more often and set the exit value correctly, then prove it with a deliberate close-and-relaunch test.
What Is SCORM Suspend Data and What Does It Store?
Suspend data is a block of text the course writes to the LMS and reads back on the next launch. The LMS stores it without trying to understand it. Only the course knows what the string means. In SCORM 1.2 and SCORM 2004 the field has the same name, cmi.suspend_data.

Typical contents include:
- Which slides the learner has visited, so the menu can show progress correctly.
- Values of course variables, such as a chosen job role or a running score.
- Quiz answers given so far, and which questions were randomly drawn from a bank.
- States of individual interactions, such as which tabs or accordion panels were opened.
Suspend data is not the bookmark. SCORM has a separate, smaller field for the learner's position: cmi.core.lesson_location in SCORM 1.2 and cmi.location in SCORM 2004. Many courses use both, a bookmark for "which slide" and suspend data for "what state was everything in".
There is one more field that matters for resume: entry. On launch, the course reads cmi.core.entry (1.2) or cmi.entry (2004). A value of "resume" means the LMS is continuing a suspended attempt; "ab-initio" (spelled "ab_initio" in SCORM 2004) means a fresh start. A course that ignores this value cannot know whether it should restore anything.
How Big Can Suspend Data Get in SCORM 1.2 and 2004?
The limits depend on the version and edition. The SCORM run-time reference gives them as follows:
| Version | cmi.suspend_data limit | Bookmark field limit |
|---|---|---|
| SCORM 1.2 | 4,096 characters | 255 characters |
| SCORM 2004 2nd Edition | 4,000 characters | 1,000 characters |
| SCORM 2004 3rd and 4th Editions | 64,000 characters | 1,000 characters |
These are the sizes the specifications let a course count on. Some LMS platforms accept more, but a course that relies on extra room is betting on one system's generosity. If the package moves to another LMS, the bet can fail.
It is easier than you might think to exceed 4,096 characters. As a made-up example, take a course with 120 slides that records each slide's state as a short entry of about 30 characters, which comes to 3,600 characters. Add 20 quiz answers at about 60 characters each, which is 1,200 more, and the total reaches 4,800. The course has passed the SCORM 1.2 limit before counting a single variable.
Notice that the limit is counted in characters, not in slides or minutes. Two courses of identical length can use very different amounts of space, depending on how much state the authoring tool records for each slide and interaction. The only reliable way to know where a particular course stands is to measure the value it actually sends.
Many authoring tools compress suspend data to make room. Compression helps, but it does not change the ceiling, and long courses with many interactions can still reach it.
Why Do Courses Lose Their Place When Learners Return?
There are a handful of usual causes, and they leave different fingerprints:
- The data exceeded the limit. The LMS rejected the SetValue call or cut the string short, and the course could not read the damaged string on relaunch.
- The exit value was wrong, because the course did not set the exit field to "suspend", so the LMS treated the attempt as finished and started a new one.
- The final commit never happened, so the data was set but not saved before the window closed.
- The LMS starts a new attempt on every launch, because of a course or enrolment setting.
- The course was replaced in the LMS, and saved data that refers to old slide IDs no longer matches the new version.
- The course ignores entry, or its resume prompt was switched off at publish.
The same causes can also wipe status. A learner who reached completion in one session may show as incomplete if that session's data never arrived. For that side of the reporting puzzle, see how completion status vs success status works.
How Do Exit Settings and Browser Closing Affect Resume?
Exit settings decide whether the LMS treats the visit as a pause or as an ending, and browser closing decides whether the course gets the chance to say which. Before closing, a course should set the exit field and then end the session properly.
- SCORM 1.2: set cmi.core.exit to "suspend", call LMSCommit, then LMSFinish.
- SCORM 2004: set cmi.exit to "suspend", call Commit, then Terminate.
With exit left empty, or set to "logout" or "normal", many LMS platforms will close the attempt. The next launch then starts fresh, which is correct behaviour for a finished course and the wrong behaviour for one that is halfway done.
Browser closing is the harder problem to solve. When a learner clicks the browser's own close button, the course gets only a brief moment to run code, and modern browsers restrict what a closing page can do, including some kinds of network requests. A course that saves everything only at the very end can lose the lot. The defence is to commit at meaningful moments, such as after each slide change or quiz submission, so that closing the window loses at most a few seconds of progress.
How Can You Shrink What a Course Needs to Save?
Save what you need to restore the learner's experience, and nothing more, because every unnecessary character brings the course closer to the limit. These changes usually make the biggest difference:
- For a linear course, store the furthest slide reached instead of a visited flag for every slide.
- Store question IDs from a bank, not the question text.
- Use short keys and codes instead of long descriptive names.
- Stop tracking states of purely decorative interactions.
- Avoid saving free-text answers in suspend data. If they matter, record them as interactions instead.
- Break a very long course into shorter modules, each with its own allowance.
Switching to SCORM 2004 3rd or 4th Edition is the other option, raising the ceiling from 4,096 to 64,000 characters. The comparison of SCORM 1.2 and 2004 sets out what else changes when you switch. Confirm with your LMS administrator that the platform supports the newer version well before you republish a whole library.
How Do You Prove the Fix Works Before Learners Notice?
Recreate the failure on purpose, then confirm it no longer happens. A reliable test routine looks like this:
- Launch the course in a test tool or LMS sandbox that logs SCORM API calls.
- Take the longest path through the course: every branch, every optional slide, every quiz question.
- Stop about two thirds of the way through.
- Check the log: note the length of the last suspend data value and confirm the exit value is "suspend".
- Close the window with the browser's own close button, not the course's exit button.
- Relaunch and confirm that entry is "resume" and the course returns to the right slide with answers intact.
- Repeat in a second browser, and once more in the real LMS.
The log also reveals truncation: if the value the course sent is longer than the value the LMS returns on the next GetValue call, the LMS cut it short. Likewise, if SetValue returns "false" and the error code is not zero, the LMS refused the value outright. Either finding points straight at cause one in the list above, and the remedy is to reduce what the course saves rather than to adjust LMS settings.
Step four is the one people skip. A course that sits at 3,900 characters on your test path is one more interaction away from failing. Treat a number close to the limit as a warning, not a pass.
A Course That Resumes Properly Is One That Saves Less, Not More
SCORM suspend data is a small, shared notebook between a course and an LMS. When resume fails, the notebook was too full, never saved, saved under the wrong exit value, or never read back on relaunch. Each of those has a specific, testable fix.
Keep the saved state lean, commit at sensible points, set the exit field to suspend when the learner leaves mid-course, and test with a hard close on the longest path. Do that before release, and the "it made me start again" message becomes rare enough to be interesting.
