Problem Diagnosis
The patent attorney’s core anxiety is not the office action on the desk but the invalidation trial five years after registration. A self-compiled database of 1,365 Korean Patent Tribunal (KIPTAB) invalidation actions—1,168 decided on the merits—shows a cumulative acceptance rate of 59.8% (Source 2). That figure is nearly double the roughly 38% rate for ex parte appeals against examiners, making invalidation the highest-risk contested proceeding in the Korean patent system. For the drafting attorney, the number means this: a specification and claim set that passes pre-grant examination still faces a better-than-even chance of being partially or fully struck down when challenged later.
The bottleneck is not merely the rate but its composition. Inventive step (Patent Act §29(2)) is asserted in 80.2% of merits cases, and when pleaded alone it succeeds 59.7% of the time (Source 2). Insufficiency of disclosure—lack of enablement under §42(3)—carries a standalone acceptance rate of 75% (Source 2). These are not marginal risks; they are the two largest levers a challenger pulls. A Korean patent attorney who treats the specification as a prosecution document rather than a litigation-ready asset is building a portfolio with a structural vulnerability that the data says will be exploited.
Analysis
The acceptance rate has not been static. From a high of 66.7% in 2020, the annual figure fell to 51.2% in 2025—a 15.5-percentage-point decline over five years (Source 4). The drop is not a straight line: 2023 saw a sharp −7.7 pp move to 53.1%, followed by a 4.0 pp rebound in 2024 and a further −5.9 pp in 2025. The 2020–2021 average of 65.85% versus the 2024–2025 average of 54.15% yields an 11.7 pp gap, confirming the trend is real even after smoothing single-year noise (Source 4).
What changed? The data cannot prove causation, but three concurrent shifts are visible. First, the KIPTAB panels appear to be applying a stricter inventive-step standard that demands more than a mosaic of known elements. The Supreme Court of Korea has repeatedly held that obviousness requires a motivation to combine with a reasonable expectation of success, not merely that each component was individually known (see, e.g., Supreme Court decisions 2008hu3377, 2017hu2543, cited in Source 8). When that doctrine percolates down to the tribunal level—typically with a two-to-three-year lag—the acceptance rate drops. The 2022–2023 inflection aligns with this timeline.
Second, the mix of cases has shifted. The 2020–2021 period likely saw a higher proportion of “easy” invalidation targets; as those cleared, the residual pool became harder. Simultaneously, patentees are investing more in pre-dispute defensive analysis, producing answer briefs that deploy hindsight prohibition arguments more effectively (Source 4). A case where an 11-document combination attack was defeated on hindsight grounds illustrates the point (Source 4).
Third, the sample size contracted. Merits decisions fell from 222–257 per year in 2020–2021 to 148–177 in 2022–2025, increasing the statistical noise around any single year’s rate (Source 4). But the 15.5 pp gap between the high and low points exceeds the ±7–8 pp confidence interval, so the long-term direction is reliable.
The appellate layer adds further pressure. Of 664 invalidation decisions appealed to the Patent Court, 197 were reversed—a 34.6% cancellation rate (Source 2). At the Supreme Court, the overall patent-field reversal rate is 41.0% (Source 2). A challenger who loses at the tribunal still has a one-in-three chance of overturning the decision at the Patent Court, and a patentee who wins at the tribunal faces a similar risk of reversal on appeal. The three-instance system is not a rubber stamp; it is a re-examination engine.
Strategic Implication
The data compels a shift in drafting and portfolio management. For the Korean patent attorney drafting a specification today, the five-year horizon is the relevant timeframe. The acceptance rate in 2025 was 51.2%; the rate that will apply to a 2026 filing when challenged in 2030–2031 will be shaped by the same doctrinal tightening that has already taken 15.5 pp off the table. The response is not to draft narrower claims—that cedes scope—but to build a specification that preempts the two most common attacks.
Action 1: Pre-load the specification with a motivation-to-combine firewall. Inventive step attacks succeed when the challenger can point to a clear suggestion, teaching, or motivation in the prior art to combine references. A specification that explicitly identifies the technical problem the prior art failed to solve, the specific obstacle that prevented the combination, and the unexpected result that flows from the claimed solution makes that combination argument harder. This is not a prosecution tactic; it is a post-grant defense built at the drafting stage. Korean filers should treat the “Effect of the Invention” section not as a formality but as the evidentiary anchor for a future obviousness defense.
Action 2: Run a pre-filing enablement audit against the §42(3) 75% acceptance rate. The standalone acceptance rate for lack of enablement is 75%—higher than any other single ground (Source 2). Before filing, ask: can a person of ordinary skill in the art reproduce the full breadth of the claimed invention without undue experimentation? If the answer requires a specific temperature, a particular catalyst, or a non-obvious parameter range, that data must be in the specification. Korean practice under Patent Act §42(3) requires the description to be “clear and detailed” enough for reproduction; a specification that omits working examples or relies on post-filing data is a target. The 30-minute investment in adding a detailed embodiment paragraph is a hedge against a 75% probability of losing the entire patent later.
Action 3: Structure the claim set for partial survival. The KIPTAB can invalidate claims individually (Patent Act §133(1)). A patent with a single broad independent claim and no fallback positions is a binary bet. Draft dependent claims that capture commercially meaningful sub-combinations, parameter ranges, and specific embodiments. If the broad claim falls, the dependent claim survives and can still cover the competitor’s product. This is not about claim count; it is about claim architecture. Korean filers in the chemical and pharmaceutical sectors—where selection inventions and Markush claims are common—should note that the Supreme Court has clarified that a difference in technical constitution that goes beyond a mere addition of known art is not identical, even if the difference is within the ordinary skill level (Source 8, citing 2017hu2369). A well-drafted dependent claim that captures a specific sub-range or a particular salt form can survive even when the genus claim is invalidated.
Action 4: For Korean challengers, time the attack to the current tribunal’s acceptance rate, not the historical average. The 59.8% cumulative average masks a 51.2% current-year rate. A challenger filing in 2026 should model its probability of success on the 2024–2025 average of roughly 54%, not the 2020 high of 66.7%. This means selecting the strongest two or three grounds—typically inventive step plus either novelty or enablement—rather than pleading every conceivable defect. The data shows that a focused attack on inventive step alone succeeds at 59.7%, while adding a weak second ground can dilute the narrative without improving the odds (Source 2).
Action 5: Monitor the appellate pipeline when deciding whether to settle. A 34.6% Patent Court cancellation rate and a 41.0% Supreme Court reversal rate mean that a tribunal win is not final. Korean parties on both sides should factor appellate risk into settlement valuations. A patentee who wins at the tribunal should not assume the patent is secure; a challenger who loses should not assume the case is over. The three-instance system creates a 12–24-month window of uncertainty that can be exploited in negotiation.
The Korean patent system is not becoming more hostile to patentees uniformly—it is becoming more demanding of the specification. The 15.5 pp decline in the invalidation acceptance rate over five years is a signal that the tribunal is reading the specification more rigorously. The attorney who drafts with that rigor in mind, and the IP manager who allocates the extra hour to pre-filing review, is not avoiding a cost—they are buying an option on validity that the data says is worth exercising.
