Enigma Machine Simulator Online

TL;DR: the same settings encode and decode

To decode an Enigma message, enter its ciphertext with the sender’s exact settings and original start positions. Try Enigma I, reflector B, rotors I–II–III, rings AAA, start AAA, and an empty plugboard: AAAAA → BDZGO. Reset to AAA and enter BDZGO to recover AAAAA. The default training message uses a plugboard; load “Five A letters” to try this check directly.

This online Enigma machine supports Enigma I with rotors I–V, and M3 with rotors I–VIII, plus wide reflectors B and C. Both use three moving rotors. You need the sender’s known settings: this simulator does not recover unknown keys or simulate the four-rotor M4.

How to use the simulator

1. Model: choose Enigma I or M3 to select the available rotor set.
2. Reflector: match B or C. This setting changes the return path.
3. Rotors: select three different rotors in left-to-right order; the rightmost rotor is the fast one.
4. Rings: enter three letters. Convert numbered settings using 01 = A, 02 = B, through 26 = Z; for example 02–21–12 becomes BUL.
5. Start positions: enter the three window letters shown before the first keypress.
6. Plugboard: enter space-separated pairs such as AB CD, with each letter used once, or leave it blank.

Choose Encrypt or Decrypt, then type or paste your message. Each edit restarts the entire calculation from your selected windows. Swap puts the output into the input without changing those starting settings. Copy and Download use the displayed result.

Only ASCII letters advance the rotors. Spaces, digits and punctuation stay in place; they are presentation conveniences. Group Letters produces uppercase A–Z in groups of five. For a historical intercept, follow its procedure for the indicator and message key before entering the body; see the Py-Enigma communication example.

Simulator verification

Computed on October 29, 2026: 7 of 7 checks pass. Expected text comes from the linked references or the default training example; Actual is calculated by this simulator. Every row includes the full starting settings and input.

A historical message: Scharnhorst

Stefan Krah’s published Scharnhorst break gives the adjusted settings used here: reflector B, III–VI–VIII, rings AHM, message key UZV, plugboard AN EZ HK IJ LR MQ OT PV SW UX. Its first 20 ciphertext letters, YKAENZAPMSCHZBFOCUVM, decode to STEUEREJTANAFJORDJAN. Load “Scharnhorst message excerpt” to reproduce it; the source supplies the complete message and its provenance.

Measured properties: 10,000 letters

0 self-encryptions; 10,000 of 10,000 letters recovered (100%). All 20 of 20 messages round-trip exactly when the original start positions are restored.

The reproducible sample uses xorshift32 with seed 0x454e4947: 20 messages of 500 letters, both models, both reflectors, varied rotor orders, rings and starts, and 0–13 plug pairs. These measurements describe this sample. Download the seven verification results and the sample’s settings, hashes and totals (JSON).

The seven verification rows include their inputs and outputs. The 20 sampled messages are represented by SHA-256 hashes in the JSON. To regenerate those messages and reproduce the totals, download the version 1 generator and source dependencies (ZIP). Extract it, then run node --experimental-strip-types verify.mjs with Node.js 22.6 or later. No package installation or network access is needed; the included README explains how to regenerate both JSON files.

Double-stepping in action

With I–II–III, ring settings AAA and starting windows ADU, the three keypresses below produce ADU → ADV → AEW → BFX. The notch is checked before movement. Rotor III turns its neighbor when leaving V; rotor II turns its neighbor when leaving E. This makes the middle rotor move on two successive presses, as explained in Rijmenants’ stepping mechanism.

The ring shifts the wiring relative to the visible alphabet. Since the notch belongs to that alphabet ring, changing rings does not change these turnover window letters. M3 rotors VI, VII and VIII each turn over at both Z and M. Load “Double stepping: ADU” and enter one, two, then three A letters to watch the final windows.

How Enigma encrypts

A keypress advances the rotors before the electrical signal passes through them. The return journey uses each rotor’s inverse wiring, making the complete path reciprocal. The Crypto Museum circuit explanation shows this route through the machine.

Why Enigma was breakable

A guessed fragment, or crib, can be aligned with ciphertext. Because Enigma never maps a letter to itself, any alignment containing such a match can be rejected immediately. That narrows a search; it does not reveal the key by itself. Compare letter frequency analysis: moving rotor substitutions do not preserve a simple one-to-one frequency mapping.

Marian Rejewski reconstructed Enigma’s wiring in the Polish Cipher Bureau’s early work; Jerzy Różycki and Henryk Zygalski also developed methods for recovering keys. Crypto Museum’s history traces that breakthrough and the sharing of Polish knowledge with Britain and France.

At Bletchley Park, Alan Turing’s bombe design used cribs to test candidate settings; Gordon Welchman’s diagonal board improved that search. Crypto Museum’s bombe account distinguishes this approach from the earlier Polish bomba. Mathematical structure, message procedures and probable plaintext all mattered.

Frequently asked questions

Explore related ciphers

Unsure which cipher you have? Start with the cipher identifier. Compare changing alphabets in Vigenère, paired-letter substitution in Playfair, or fractionation and transposition in ADFGVX.