The Death Ledger

Every record on this site tracks arrivals. First appearances are the franchise: who showed up, in which issue, on what date, and whether anyone disputes it. But the same structured indexing that dates the arrivals also flags the departures, and nobody ever ranks those.

So this article does. Across the 1,537,367 character appearances in this register, exactly 4,450 carry a death flag outside of flashbacks, spread across 2,870 distinct characters. Hold that against the size of the record and the first finding is already on the table: recorded death in comics is rare, about three appearances in every thousand. The second finding is the famous one, except now it has a number attached. The dead come back, two thirds of the time, on a median of two years.

Line the deaths up by decade, per thousand appearances so the size of each era's output cannot distort it.

Decade	Recorded deaths	Per 1,000 appearances
1930s	11	3.98
1940s	18	0.56
1950s	7	0.17
1960s	69	1.01
1970s	258	2.50
1980s	718	4.11
1990s	989	4.63
2000s	857	3.59
2010s	927	2.83
2020s	606	2.03

The curve rises from almost nothing to a 1990s peak and then falls for thirty years, and the obvious reading is that the industry got more violent and then calmed down. That reading is probably part of the truth and definitely not all of it, so here is the caution before the conclusions.

The 1950s line records seven deaths in an entire decade. The temptation is to call that the Comics Code, and the Code is certainly in there: it arrived in late 1954 and it did restrict what could be shown. But the first half of that decade is EC horror and crime, books that killed people with enthusiasm and were dragged in front of a Senate subcommittee for it, which is the subject of its own article. Seven is not a believable count of deaths printed in the 1950s. It is a believable count of deaths somebody has flagged in the 1950s.

That distinction runs through the whole table. The death flag is applied by cataloguers working story by story, and that work is far more thorough on recent and heavily collected material than on seventy-year-old anthologies. Normalising per thousand appearances controls for how much was published. It does not control for how carefully it was indexed.

So read the decade curve as editorial practice and indexing density moving together, with no clean way to separate them. What survives that caveat intact is the shape of the recent end, where indexing is dense and comparable throughout: the rate peaks in the 1990s, and the 2020s run at less than half that peak. The nineties were deadlier than the decades on either side of them, and the correction worth making is that they were only about thirteen percent deadlier than the eighties, not the order of magnitude the reputation implies.

Here is the number this article exists for. Of the 2,870 characters with a recorded death, 1,910 appear again in the register afterwards. That is 66.6 percent. The median gap between a character's first death and their next appearance is two years. More than half, 53.7 percent, are back within two years, and two thirds are back within five.

I want to be precise about what the register is measuring, because it is both less and more than resurrection. An appearance after death means a newly indexed, non-flashback story carrying the character, and that includes reprints of older material, so some of these returns are publishers republishing the character rather than writers reviving them.

I considered filtering that out and decided it is the truer measure left alone. In this medium, dying does not even keep a character out of print. Whether the return is a resurrection storyline or an archive edition, the commercial fact is identical: two years after a death, the character is back on a page you can buy.

340 characters in the register have died three or more times. 74 have died five or more times.

The raw leaderboard of most-recorded deaths is one of those boards this site prints with a warning label, the same way the raw board in The Haunted Leaderboard needed one, because the top of it is not Jean Grey. It is gag characters who die as a running joke, humour-strip casualties, the Stay-Puft Marshmallow Man perishing essentially every time he manifests, Galactus's herald Air-Walker, whose entire function is dying with dignity at regular intervals, and at least one entry that appears to be the Joker as indexed in Brazilian editions, his deaths counted again in Portuguese.

The register counts recorded tellings, not souls, and reprinted deaths count as new records the same way reprinted debuts do. That is not a flaw to hide. It is what an honest index of 1.5 million appearances looks like before an editor smooths it, and the aggregate numbers above it, 340 three-timers and 74 five-timers, survive the caveat intact.

Flip the question. Which characters with real careers, twenty or more appearances, died and never appeared again?

I expected this board to be the article's monument, the Uncle Ben table. The register handed back something better, because the board is almost entirely deaths from 2021 onward. Blink, with 462 prior appearances. Sabrina Spellman, 313 appearances. Doctor Solar, the Gold Key atomic hero. Jann of the Jungle, an Atlas heroine from the 1950s, killed in a modern book.

These deaths have not stuck. They are simply too recent to have been reversed yet, and the register is honest enough to show it: the only character on the whole board with more than forty appearances who has stayed dead for over a decade is Brideshead, gone since 2013.

Read that again, because it is the strongest sentence in this ledger. Across a century of publishing and 1.5 million indexed appearances, the record cannot produce a single major, long-running character who died more than fifteen years ago and stayed dead. In this medium, staying dead mostly means having died recently.

Rank single issues by distinct characters killed and the top of the raw board belongs to omnibuses and archive editions: a Heroes Reborn omnibus logging 40 deaths, a Savage Dragon archive 34, an Infinity Gauntlet omnibus 26. That is the revolving-door artifact at full scale, because a collection inherits every death it reprints.

Strip the collections away and the genuine crown lands exactly where the decade table said it should. The deadliest original comic book in the register is Deathmate Black, the 1993 Image and Valiant crossover, 28 characters killed in one black-covered issue, published within months of the peak of the per-thousand death rate.

The nineties did not just have the highest recorded body count per page. They produced the single bloodiest issue ever indexed, and it was a speculator-era crossover event, which is about as on the nose as the record ever gets.

Deathmate Black on eBay ›

The rest of the original-issue board reads like a syllabus: West Coast Avengers #26 with 20 deaths in 1987, Savage Dragon #103 with 19, Erik Larsen killing as freely as he creates, and Marvel Super Heroes Secret Wars #11 with 18, late in the 1985 event.

One last cut. 384 characters in the register die in their very first recorded appearance. Their debut and their death are the same entry, one issue, in and out.

Most are exactly what you would guess: henchmen, victims, one-scene casualties indexed by diligent cataloguers, the medium's redshirt census. But as a class they are the purest expression of what the death flag is for. The record does not decide who mattered. It writes down who appeared and who died, and sometimes that is the same line.

Three sentences. Recorded death in comics is rare, three appearances in a thousand, and the rate has halved since the nineties. Death in comics is temporary, two thirds of the dead return at a median of two years, and the register cannot find one major character dead more than fifteen years. And the recorded peak, in rate and in the single bloodiest issue, falls in exactly the decade everyone blames, black covers and all.

The fourth sentence is the one this site insists on. Every figure above measures what has been written down, not what happened, and the gap between those two things is widest in the oldest material. The 1950s row is the proof.

Every death record in this ledger sits on an issue page you can check, the dated first appearances live in the register, the chronology games the dates play are documented in The Cover-Date Lie, and the contested arrivals are in the disputes file. If the record shows a death that never happened, or missed one that did, the corrections address in the footer reaches a person and the ledger gets amended. That is the whole method. Nobody stays dead here, but everybody stays on file.