CSV to ICS

About this tool

Calendars import .ics files; spreadsheets export CSV. This tool is the missing step between them. Paste a table of events — one row per event, with a header row naming the columns — and it hands back a standards-compliant iCalendar document you can save as schedule.ics and import into Google Calendar, Outlook, Apple Calendar, Thunderbird or anything else that speaks iCalendar.

The conversion runs as WebAssembly inside this page, so your schedule never leaves the browser tab. The output is deterministic too: the same CSV always produces the same bytes, which keeps re-imports and version-controlled calendar files clean.

A worked example

Paste this, leave every option alone:

title,start,end,location
Team sync,2026-07-24 09:00,2026-07-24 09:30,Room 2
Conference,2026-07-27,2026-07-29,Berlin

You get back:

BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//gizza-ai//csv-to-ics//EN
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:[email protected]
DTSTAMP:19700101T000000Z
DTSTART:20260724T090000
DTEND:20260724T093000
SUMMARY:Team sync
LOCATION:Room 2
END:VEVENT
BEGIN:VEVENT
UID:[email protected]
DTSTAMP:19700101T000000Z
DTSTART;VALUE=DATE:20260727
DTEND;VALUE=DATE:20260730
SUMMARY:Conference
LOCATION:Berlin
END:VEVENT
END:VCALENDAR

The first row had a time, so it became a half-hour timed event. The second had bare dates, so it became a three-day all-day event — note the DTEND of the 30th, because iCalendar all-day ends are exclusive and the tool adds that day for you.

Columns it recognises

Header names are matched without regard to case, spaces, hyphens or underscores, so Start Date, start_date and START-DATE are the same column. Anything unrecognised is ignored rather than rejected.

• Title (required) — title, summary, name or event
• Start (required) — start, start_date, begins or date
• End — end, end_date or ends
• Duration — duration_minutes or duration, in whole minutes
• Description — description, details or notes
• Location — location or place
• UID — uid or id
• All-day — all_day or all-day, true for true, yes, y, 1, x or t

Dates, times and lengths

Dates are written as 2026-07-24, 2026-07-24 09:00 or 2026-07-24T09:00, with optional seconds. A start with no time — or a truthy all-day cell — makes an all-day event written as VALUE=DATE. Anything else is a timed event, and it ends at its end cell, else its duration_minutes cell, else the default event length (60 minutes unless you move the slider).

The timezone choice decides how timed events are anchored. Floating writes DTSTART:20260724T090000 with no zone marker, so 09:00 shows as 09:00 wherever the event is opened — the right choice for a class timetable or a personal schedule. UTC treats the pasted times as already being UTC and writes them with the trailing Z, so each calendar converts them into its own local time — the right choice for a release plan or a call across several countries. All-day events carry no time either way.

Limits and edge cases

• Quoted cells are read as one value, so "Lunch, with team" keeps its comma, and commas, semicolons, backslashes and newlines are escaped in the output the way RFC 5545 requires.
• Lines longer than 75 octets are folded with a leading space on the continuation, so long descriptions import intact instead of being truncated by a strict parser.
• Up to 5000 rows per run. Rows that are entirely blank are skipped.
• A row with an unreadable date, a duration that is not a whole number of minutes, a missing title or start, or an end that falls before its start is reported by row number — the file is not written with a broken event in it.
• Event IDs come from the uid column when you have one. Without it, an ID is generated from the title and the row's position, which stays stable across runs, so re-importing the same file updates the events instead of duplicating them.

FAQ