If you searched “how to restart a dnd campaign after a break,” here is the promise up front: the campaign is almost certainly still alive, the way back in takes one short meeting and one good session, and the biggest mistake you can make is the one most DMs reach for first. I’ll give you the whole restart, in order.

The mistake: trying to recap your way back in

The instinct is to treat the break as a memory problem. Write a heroic summary of sessions one through eleven, read it at the table, and resume as if the gap never happened. I’ve watched this fail at my own table and I understand exactly why it’s tempting: it feels responsible. You’re the keeper of the story, the story got dropped, so you pick it up and hand it back.

But a three-month break does something a three-week gap doesn’t. Across a normal gap, players forget logistics and keep stakes, which is why the salvage protocol for a missed session works at all. Across a whole summer, the stakes go too. Your rogue doesn’t just forget the smuggler’s name; she forgets being angry at him. The emotional threads that made the campaign feel urgent have gone slack, and no summary, however well written, can re-tension them by being read aloud. A recap assumes an audience that still cares and has merely misplaced the details. After a long break you have the opposite: people who remember they loved this, in the abstract, and feel almost nothing about the duke, the debt, or the door you left dramatically unopened.

So treat the restart as what it actually is: a re-onboarding. You are re-recruiting five people into a story they already said yes to once, which is a much easier sell than the first time, as long as you run it like a sell.

Step one: a twenty-minute session zero point five

Before you prep a single scene, get the table together for twenty minutes. Voice call is fine. This is a miniature version of session zero, and it answers two questions.

“Are we still these characters?” Ask each player, directly: are you still excited to play this person? Three months is long enough for the answer to change, and it’s enormously better to hear “honestly, I was getting bored of the barbarian” now than to watch someone sleepwalk through four sessions. Offer the cheap fixes out loud: a retooled subclass, a new personal hook, or in the extreme case a new character written into the story. A player who chose their character again five minutes ago plays differently than one who got handed their old sheet.

“What do we still owe each other?” This is the scheduling conversation, and it’s the actual reason the campaign stopped. Campaigns die of calendars, and a campaign that has already died of one once will do it again unless something changes. Re-pick the slot with calendars open. Re-agree the quorum rule. If the old slot was the problem, say so and move it. Skipping this step because it’s awkward is how you end up writing this same group-chat message in January.

Twenty minutes, both questions, done. Do it days before the first session back, so the anticipation has somewhere to build.

Step two: canonize a “story so far,” don’t perform one

You still need the table to share a version of what happened, and here is the arguable position: write it as two paragraphs, send it in the chat before game night, and declare it the official record. Two paragraphs, maximum, and the ceiling is doing the work: whatever survives the compression is, by definition, what the campaign is actually about, and whatever gets cut is stuff the story can live without.

Write the first paragraph as the situation: who the party is, what they want, and the one debt or threat that defines the current moment. Write the second as the cliff edge: where they were standing when the music stopped, and the question the next scene has to answer. If you have real records, mine them for the details your memory has sanded off. This is where months-old session recaps stop being an archive and start being a rescue: when we built Loracle’s campaign memory, the case we kept designing for was exactly this one, a table coming back after a gap long enough that even the DM needs the record. The recordings and recaps don’t fade in June. You can ask the campaign what the hag’s deal actually was, in the words the table used, and put the true version in your two paragraphs instead of the version you’d have sincerely misremembered.

Then, at the table, read it once and rule that it’s canon. Somebody will remember events differently. Across three months, everyone’s memory is reconstruction; the written version wins, the table nods, and you’ve spent ninety seconds on what could have been a half-hour continuity hearing.

Step three: restart on a hard hook, not an admin session

The first session back must be a real session, and it should be the most playable one you’ve run all year. There’s a strong pull toward an easing-in night: level up the characters, redo the shopping, “get our feet back under us.” Resist it with everything you have. An admin session confirms every quiet doubt in the room that the campaign is a chore you all used to do. Handle the bookkeeping asynchronously in the chat during the week, and spend the actual slot proving the game is alive.

Open in motion. First scene, somebody has to make a decision with a consequence hanging off it. The door you left unopened in June gets kicked in by someone else. The debt gets called. The NPC the party trusts sends word that things went badly while they were away, because time passed in the world too, and a world that moved without the party is instantly more real than one that waited politely under a tarp. You want a player leaning in by minute ten, texting the group something in character by the next morning. That text is the campaign restarting; everything before it was setup.

And when the night ends, end it the way that would have saved you in June: on a live question, with the last five minutes spent booking the next session before anyone logs off. The restart is finished the moment session thirteen happens on schedule, and no sooner.

The campaign was never dead

Here’s the exit thought, and I believe it literally: a campaign with no scheduled session is paused, and a paused campaign holds its value for a shockingly long time. The characters keep meaning something to the people who played them. The goose oath stays funny. What decays is the connective tissue, the names and debts and unopened doors, and that decay is only fatal if nobody kept a record.

So keep one, starting now, whichever way you keep it. Loracle does it by sitting in the Discord call and turning each session into a recap and a searchable campaign memory, which means the next break, and there will be a next break, because you are adults with jobs, costs you nothing but the time off. The story waits, in full, exactly where you left it. Three months from now, “we should play again” can be a plan instead of a wish.

This is what a D&D game across three time zones looks like when it’s working. The college group scattered, or the forum friends never lived in the same country to begin with, and now every session is simultaneously a Saturday night, a late night, and a Sunday breakfast. If you searched “how to run D&D in different time zones” because your group chat has spent two weeks failing to schedule session one, here is the honest answer: the game is absolutely runnable, people run campaigns like this for years, and almost everything that kills them is a logistics mistake with a known fix. The fixes below are the ones that matter, roughly in the order groups get them wrong.

Stop looking for a time that works for everyone

There is no fair time. Spin a globe with players in New York, London, and Sydney on it and every slot you land on is somebody’s midnight or somebody’s dawn. The groups that die spend months refusing to accept this. They run a fresh availability poll every week, hoping this time the grid will produce a slot nobody hates, and every week the poll itself burns a little more of everyone’s enthusiasm. Scheduling by consensus is a tax you pay weekly, forever.

Pay it once instead. Find the least-bad time, give the person who got the worst end of it a genuine veto, and then write the result down as law: same slot, every week, permanently. This belongs in session zero alongside tone and safety tools, because at long distance the scheduling contract carries more of the campaign’s weight than anything else you agree on. Most campaign deaths are scheduling deaths, and a standing time is the strongest medicine there is. A recurring slot becomes a habit, and habits survive busy months; a weekly negotiation has to win an argument every single week, and eventually it loses one.

Anchor the game to one named city

“6 pm Eastern” feels precise. It is a trap with a fuse on it, and the fuse is daylight saving time. The United States and Europe change their clocks on different weekends in spring and autumn, so for a few weeks a year the gap between New York and London is five hours in one direction. Australia changes in the opposite season entirely. Twice a year, a game scheduled by offset quietly detonates: two players arrive an hour apart, both of them certain they’re on time, both of them right.

The fix costs nothing. The game lives in one city’s civil time, named out loud: “we play at 6 pm New York time, whatever that happens to be where you are.” One person owns a recurring calendar invite set to that city’s time zone, everyone else accepts it, and the software does the conversion forever. When the clocks change, the invite moves everyone together. The DM’s most important prep artifact for a distributed table is a calendar event with the right time zone attached. That’s the whole trick, and groups still get it wrong for years.

Run shorter sessions than you think you want

The four-hour session is a same-living-room luxury. In your three-city game, hour four of a 6 pm New York start is 3 am in London. Your wizard was sharp for the first two hours, quiet for the third, and by the fourth she’s a camera-off presence who answers direct questions after a delay. Push a long session on a distributed table and the back half is played by fewer people than the front half, whatever the attendance list says.

Two to two and a half hours, weekly, is the shape that works. It fits inside a school-night evening and a weekend morning at the same time, so nobody has to be a hero to attend. It keeps the whole table inside the hours where they’re actually good. And a shorter session concentrates play in a way that’s easy to underestimate: when everyone knows the window is two hours, the shopping montage gets snappy and the table finds the plot faster. Ending on time, on purpose, is its own craft skill, and it matters double here; I wrote up how to land an ending so that next week starts itself.

Spend live hours only on things that need to be live

Once you accept the two-hour window, the arithmetic gets strict. Live, synchronous time is the scarcest resource your game has, and a lot of what fills a typical session doesn’t need it. Shopping trips. Downtime projects. The letter your bard wants to send to the duchess. Two players having the quiet character conversation they’ve been circling for weeks. All of that plays beautifully in a side channel on Discord, typed across days, at whatever hour suits each writer.

Move it there. Reserve the live session for the things that genuinely require six people in one moment: fights, revelations, arguments about the plan, and any decision the whole party has to own. The async channel does double duty, too. Your Sydney player, whose live slot is breakfast, gets a version of the game where the hour is always right for him, and roleplay that would have been rushed at the table gets room to breathe. Some of the best scenes in a long-distance campaign happen on a Wednesday, in text, between two people who have never been awake at the same time.

Design for absence, because absence is coming

Run any game long enough and people miss sessions. Run one across three time zones and the misses multiply: the 8 am player has a family thing, the 11 pm player has an early flight, somebody’s power is out in a storm the rest of the table didn’t know was happening. Treat absence like weather. It arrives on its own schedule, it’s nobody’s fault, and a campaign built to require perfect attendance is a campaign with a countdown clock on it.

Building for weather takes three rules. The game runs at three players or more, announced up front, so one absence never cancels a session and the quorum question never has to be litigated at 5:55. An absent character fades politely into the background, and the table doesn’t puppet them through anything their player would want to play. And catching up costs the absent player nothing. That last one is where distributed tables quietly fail: when catching up means reading forty minutes of scrollback or cornering the DM for a summary, people skip it, fall out of the story, and drift out of the game without ever deciding to quit.

This is the exact problem we built Loracle around. The bot sits in your Discord voice channel, records the session, and posts a recap to the channel afterwards, so the player who slept through Saturday wakes up to the story on Sunday and walks into next week current. A distributed game leans on that harder than any other kind, because at three time zones the recap is doing quiet work every single week for somebody.

The game is worth the spreadsheet

Back to that Saturday. Your wizard is typing her spell descriptions in a hush at half past midnight. Your rogue has finished his cereal and is arguing tactics with a spoon in his hand. On paper this is a ridiculous way to spend a weekend, and every one of these people chose it, again, for the fortieth week running, because the alternative was letting the group dissolve when the distances got real.

A time-zone game asks more of you as a DM in exactly one place: logistics. Fix the slot, name the city, shorten the session, push what you can into the week, plan for the weather. None of it is glamorous. All of it is the difference between a campaign that limps through six sessions of scheduling polls and one where three cities’ worth of friends get the best two hours of their week at the same table, even when the table is an ocean wide.