Timing is everything with an information disclosure statement, and it gets strictest right at the end. Once a case is allowed, the rules for getting an IDS considered tighten, and the whole thing turns on a single date-driven certification. When that certification is even slightly off, the reference does not just get a warning. It falls outside the rule it was relying on, and getting it back in front of the examiner means reopening prosecution.
The small print after allowance
Under 37 CFR 1.97(d), an IDS filed after a Notice of Allowance but before the issue fee is paid must include two things: the fee under 37 CFR 1.17(p), and a certification under 37 CFR 1.97(e). The certification most practitioners rely on, 1.97(e)(1), states that each listed item was first cited in a communication from a foreign patent office in a counterpart application no more than three months before the IDS was filed. The alternative, 1.97(e)(2), states that no item was known to any individual with a duty of disclosure more than three months before the IDS.
Both are precise, calendar-driven statements. And that three-month clock is the entire basis on which a late IDS is accepted. If the certification does not actually hold, the IDS is not properly before the examiner, no matter that it was physically filed on time.
The certification that was a few days off
A counterpart foreign Office Action lands while a U.S. case is sitting under a Notice of Allowance. The team prepares an IDS to disclose the newly cited references and files it with the 1.97(e)(1) certification, comfortably inside what they believe is the three-month window, and pays the 1.17(p) fee. Done, or so it seems.
Weeks later, a review catches the problem: the reference had actually been cited in the counterpart more than three months before the IDS was filed. The date used for the certification was the date the foreign associate forwarded the Office Action, not the date the foreign office issued it. By the time the error surfaces, the real three-month window has closed, so the certification cannot be corrected and re-relied upon.
With the safe harbor gone, the only reliable way to get the art considered is to reopen prosecution. The case needs a Request for Continued Examination, filed with the IDS, so the examiner considers the references before the patent issues.
Cost: an avoidable RCE fee, plus the delay of reopened prosecution
The same mistake, discovered after the issue fee
Now run the same scenario, but the defective certification is not caught until later, after the issue fee has been paid and the case is on its way to grant. The three-month window is long closed, and the RCE path is no longer enough on its own, because you cannot file an RCE on an application that is already in the issue queue.
First you have to get the application back out of issue. That means a petition to withdraw the application from issue under 37 CFR 1.313(c), for the purpose of considering an RCE. Only once the petition is granted can the RCE and IDS go in.
So the same small certification error now costs a petition fee and an RCE fee, more attorney time, and a longer delay to grant, all to disclose references that could have been placed before the examiner cleanly if the certification had matched the record in the first place.
Cost: a petition fee, an RCE fee, and weeks of delay
There is a narrow relief valve here, the USPTO's Quick Path IDS (QPIDS) pilot, which can allow an IDS after issue-fee payment without a full RCE when the examiner determines no reopening is needed. But QPIDS still requires a proper certification and includes a conditional RCE that takes effect if the examiner disagrees. It does not rescue a certification that was wrong on the facts.
Why the certification is where it breaks
None of these are exotic mistakes. The certification depends on getting a specific date right and comparing it to the filing date of the IDS, under time pressure, at the very end of a case when everyone wants it closed. The date that goes into a 1.97(e)(1) certification is the date the foreign office issued the citation, not the day it reached your inbox, and mixing those up is easy. So is assuming nothing was known earlier when, somewhere in the file, it was.
Getting it right at filing
The fix is to tie the certification to the case's actual timeline before the IDS goes out, not to reconstruct it after a notice from the Office. This is what IPPrep Pro's 37 CFR 1.97 timing advisor is for: you give it the application's status and when the references were known, and it recommends the certification that actually applies, and checks the corresponding boxes on the SB/08. When the case is already allowed, it flags that you are in 1.97(d) territory and that both the certification and the fee are required, so the safe harbor you are relying on is genuinely available, not just asserted. See the IDS walkthrough for how the intake and certification fit together.
Getting the certification right at filing is a few seconds of attention. Getting it wrong, as these two cases show, is an RCE, and sometimes a petition to go with it.
This article is general information about USPTO practice, not legal advice. The scenarios are illustrative composites. Confirm current rules and fees against the MPEP and the USPTO fee schedule for any specific matter.