Banned Pokémon Card Art: What Censorship Cost

Twelve Pokémon cards had their artwork changed, recoloured or withheld between the Japanese and the English release, and Bulbapedia documents each one on its own card page. We priced all of them against the cards printed beside them. Across **605 English and Japanese card pairs** matched by name across seven Wizards of the Coast era expansions on September 23, 2026, the median Japanese ask is **1.04 times** the English one. Every one of the five largest ratios in that panel is a card whose English artwork was redrawn or recoloured. The largest is Grimer from Team Rocket, where the Japanese original asks 51.22 times the English card.

This is a price study, not an art-history one. The list of altered cards is taken whole from Bulbapedia before any price was read, and every source page is linked in the table below. Nothing on the list was selected because it was expensive.

Bulbapedia has no single page listing altered artwork, so the list was assembled from its individual card pages and its regional-differences category. Twelve physical cards carry a documented change. Two further cards, the Ho-Oh ex and Lugia ex immersive prints in the digital Pokémon TCG Pocket set Wisdom of Sea and Sky, were pulled after release on July 30, 2025 and reissued with new art on August 8, 2025; they are digital, they have no marketplace product, and no price in this study refers to them.

Two stories that circulate widely are not on the list because the sources do not support them. Koga's Ninja Trick was **not** withheld from English: the card exists as Gym Challenge 115/132 with the symbol removed, and the complaints in 1999 came from imported Japanese cards. Registeel's arm position was changed in the **video games**: the change Bulbapedia documents is to the Diamond and Pearl sprite in non-English European releases, later redrawn for Platinum, and it has nothing to do with a card. A third, the Uri Geller dispute that kept Kadabra out of print from 2003 until 2023, is real and is a withholding of a card rather than of an artwork, so it is out of scope here.

To know whether 51.22x is large you need the distribution it sits in. So we matched every priced English card in seven Wizards-era blocks to the Japanese printing of the same card, on a folded name that survives apostrophes, accents and the gender signs, and took the ratio of the two asks. 605 English names matched. 16 did not, and all 16 are named in the methodology: twelve basic Energy cards, three Unown letters and the Team Rocket secret Dark Raichu, none of which has a Japanese counterpart in the matching block.

The median pair trades within four percent of parity. 115 of the 605 pairs reach 2x, 16 reach 5x, six reach 10x and four reach 20x. The Japanese side is not uniformly dearer: it is cheaper than English on 295 of the 605 pairs, and the Gym era median is 0.63x, which is Japanese cards trading at a third off.

Sorted by ratio, the top of the panel is mostly the altered-art list. Ranks 1 to 5 are all documented changes. Ranks 6, 8 and 10 are not, and all three are the same artefact: our catalog records the English row at a plain rarity and the Japanese row at a foil one, so the ratio is partly a foil premium rather than an art premium. That is the one mechanical confound in a name match, and the next section removes it.

Restrict the panel to the **476 pairs whose two rarity strings agree on whether the card is a foil**, which drops 129: the three mismatches above and 95 other disagreements, plus 31 pairs where one side carries no rarity string at all and the class cannot be read either way. The median falls to 0.97x. The top of that controlled ranking is cleaner still: Grimer, Koga's Ninja Trick, Moo-Moo Milk, Jynx, Card-Flip Game and Ruin Wall take the first six places, and the seventh is an ordinary Base Set Trainer. Sabrina's Gaze leaves the controlled panel only because its cheapest Japanese printing is a theme-deck card our catalog records with no rarity at all; its booster printing, Challenge from the Darkness at $40.58, is a Common against a Common and would rank third at 28.78x.

Asks are what a seller wants. The sold side is thinner and more lopsided. **Two of the twelve English cards carry a recorded PSA 10 sold median at all**, against eight of the twelve on the Japanese side. An absent row here means no verified PSA 10 sale is on our tape for that product, not a sale at zero.

Grimer is the clearest case in the study. The English card asks 94 cents and has no recorded PSA 10 sale at any price, though 24 verified PSA 9 sales stand behind its $46.22 nine. The Japanese original has a $1,206.00 PSA 10 sold median standing on 25 verified sales. We do not print a population figure for the Japanese card: its registry line reads 3 gems against those 25 sales, which is proof the line is shared with other products rather than proof the card is scarce.

Where a Japanese sold median exists on both sides, the graded gap is smaller than the raw one. Misty's Tears is 5.64x on asks and 1.57x on PSA 10 sold medians, though the English figure stands on three sales and the Japanese on four, which is not enough to lean on. Sabrina's Gengar is 2.03x on PSA 10 with 6 and 26 sales behind the two numbers.

A study that only reports its confirmations is advertising. One of the twelve does not behave, and one cannot be measured here at all.

Sabrina's Gengar is the strongest counterexample. Its graveyard background was removed for English and it is the most expensive card in Gym Heroes at $465.16, sitting in the 99th percentile of that set. Its Japanese printing asks $800.00. That is 1.72x, ordinary for a Holo Rare pair, and its cheaper Gym Challenge printing ranks **385th of 605** in the panel at 0.78x. The removal of the crosses is documented and it is worth nothing in the price.

Magmortar from Dragons Exalted is the only documented redraw outside the Wizards era, and it is not in the panel because its block fails a sanity check rather than because it is inconvenient. The Japanese BW5 blocks quote a hard listing floor: their median ask is $3.00 and 60 of their 83 priced cards sit at $5.00 or below, against an English Dragons Exalted median of $0.60 with 81 of 128 cards under a dollar. A floor of that shape manufactures 10x to 15x ratios on ordinary commons, so the block measures the floor rather than the artwork. For the record, the Magmortar pair reads 71.81x, $1.15 against $82.58, and the Japanese card has a $844.89 PSA 10 sold median on 15 sales while the English one has no recorded graded sale. We would not cite that number without the block behind it.

It cannot show that the censorship caused the gap. Every one of these cards is also a card with a story attached, and a story is a demand shock whether or not a publisher caused it. The panel establishes that the gap is unusual inside the 605-pair population it sits in, and it stops there.

It cannot separate the redraw from the print run. Japanese Wizards-era print runs differ from the English ones card by card, and our catalog does not hold print figures for either side, so a rare Japanese card and a famous Japanese card look identical in this data.

Raw prices here are asks, not sales, and on a card with a handful of listings an ask is one seller's opinion. How thin the book behind Grimer at $48.15 actually is, we cannot say: our listing-depth table carries a row for 1,602 products and none of the twelve Japanese printings in this study is one of them. The graded table is the corrective, and on four of the twelve there is no PSA 10 sold median on either side.

The panel is seven blocks covering eight English expansions released between 1999 and 2001, chosen because that is where eleven of the twelve documented changes live. It is not a statement about Japanese cards generally. Our wider comparison of the two markets is Japanese vs English Pokemon Cards.

And the list is Bulbapedia's. It is a fan-maintained wiki, its own pages mark several of the motives as inference rather than an official statement, and a thirteenth altered card that nobody has written up would not be in this study.

Three observations would. A card with no documented art change reaching 20x on a rarity-agreeing pair would put an ordinary card among the three that clear that bar in the controlled subset, and the top of the ranking would stop being a list of art changes. A rebuild of the Japanese asks that moved the Grimer, Koga's Ninja Trick or Moomoo Milk quote materially would move ranks 1 to 3 directly, and none of the three carries a listing-depth row that would tell us in advance how likely that is. And a documented art change we have missed, sitting at or below the 1.04x median, would weaken the pattern in the other direction: ten of the eleven panel cases inside the top 22 of 605 is a pattern, ten of twenty would not be.

The case list is taken from Bulbapedia card pages and its Cards with regional differences category, fixed before any price was read, and each row links the page it came from. Two digital Pokémon TCG Pocket prints pulled in July 2025 are named but not priced because they have no marketplace product. The panel is built from the `cards` and `price_history` tables in production on 2026-09-23, day 20719 since epoch. That day is chosen because our daily tape holds 25 days and nine of them, 2026-09-08 to 2026-09-17, are partial dumps carrying 35,700 to 50,400 rows against 51,822 to 52,378 on the rest; 2026-09-23 is both the latest complete day and the largest at 52,378 rows. Rows are deduplicated on the TCGplayer product id before anything is counted, because 1,212 product ids are claimed by two catalog rows each and an undeduplicated join double counts them; where two rows claim one product the lexically smallest card id survives. Seven English blocks are matched to their Japanese counterparts: Base Set to Expansion Pack and Expansion Pack (No Rarity), Jungle to Pokemon Jungle, Fossil to Mystery of the Fossils, Team Rocket to Rocket Gang, Gym Heroes and Gym Challenge together to Leaders' Stadium, Challenge from the Darkness, City Gym Decks and the Japanese Gym Challenge deck, Neo Genesis to Gold Silver to a New World and the two Neo Intro Packs, and Neo Discovery to Crossing the Ruins. The English sets are pooled with their Japanese counterparts at block level rather than set to set because the two sides do not split the same cards into the same boxes: matching set against set drops 41 and 49 English Gym cards that exist perfectly well on the Japanese side. Names are matched after folding: NFD normalisation with combining marks removed, curly and straight apostrophes removed, the gender signs rewritten as m and f, parenthesised and multi-character bracketed qualifiers removed, a trailing collector-number suffix removed, the spelling impostor folded to imposter, then everything but letters and digits removed. One row per English name per block, the cheapest priced printing, against one row per Japanese name, also the cheapest; cheapest on both sides pushes every ratio toward 1 and therefore understates the outliers this study is about. 605 English names matched a Japanese one and 16 did not: Dark Raichu (base5-83), six Gym-era basic Energies (gym1-127, gym2-128, gym2-129, gym1-130, gym2-131, gym1-132), six Neo Genesis basic Energies (neo1-106 to neo1-111) and three Neo Discovery Unown letters (neo2-50 N, neo2-67 E, neo2-69 O). 36 Japanese names had no English pair. The foil control classes a rarity string containing holo, shiny, super rare or star as foil, an empty or None rarity as unknown, and anything else as plain; 476 pairs agree on a known class, 98 disagree and 31 carry an unknown on one side. Graded figures are the latest reading per card and grade in `graded_snapshots`, which are eBay sold medians with the verified sale count beside them, keyed by both card id and product id because the cron path writes one and older rows the other. Registry figures come from `population_snapshots`; where a card's PSA 10 sold count exceeds its registry gem count the line is shared with other products and no population absolute is printed for it, which is the case for the Japanese Grimer and the Japanese Magmortar. Medians are the upper-middle value on even samples. Ratios are Japanese ask divided by English ask, both TCGplayer market asks, never sold prices.