The Complete Guide to Suno Prompts & Meta Tags (2026)

Suno will happily turn a three-word prompt into a song, but the gap between “a pop song about summer” and a properly built prompt is huge. With the right tags you decide the genre, the vocal, the arrangement and even where the song ends. Without them, you get whatever the model feels like that day. This guide covers how the two input fields actually work, every structure meta tag worth using, the style tag categories that matter, five complete prompts you can adapt, and the mistakes that quietly ruin generations.

Style prompt vs. lyrics box: two fields, two jobs

In Custom mode, Suno gives you two separate inputs, and mixing them up is the single most common beginner mistake.

• The style box describes how the song sounds: genre, mood, tempo feel, vocal type, instrumentation, production character. Nothing you type here gets sung.
• The lyrics box holds the words that will be sung, plus square-bracket meta tags like [Verse] and [Chorus] that shape the arrangement.

Put “sad piano ballad” in the lyrics box and Suno will try to sing the phrase “sad piano ballad.” Put a chorus in the style box and it gets ignored. Sound goes in the style box, words go in the lyrics box, structure goes in brackets. Simple mode collapses everything into one description field and writes its own lyrics, which is fine for sketching ideas, but Custom mode is where real control lives.

Character limits have grown over the versions (the style box now takes several hundred characters, the lyrics box a few thousand), so treat the counter under each field as the source of truth rather than any number you read online.

Structure meta tags that actually work

Meta tags are labels in square brackets that you drop into the lyrics box. They are not sung (usually), and the model treats them as strong suggestions about where each section starts. Stick to the standard set; invented tags like [sick drop here] confuse the parser and occasionally get vocalized.

• [Intro] — the opening bars. Leave it empty for an instrumental open, or put a single line under it for a sung intro.
• [Verse] — the storytelling section. Suno keeps energy lower here and saves the lift for the hook.
• [Pre-Chorus] — a short build that ramps tension into the chorus. Two lines is plenty.
• [Chorus] — the hook. Paste identical text under every chorus tag; the model matches the melody far more consistently when the words match.
• [Bridge] — contrast before the final chorus. New chords, new angle on the lyric.
• [Instrumental Break] — vocals rest, instruments take over. Far more reliable than writing “(guitar solo)” as plain text, which often gets sung.
• [Outro] — winds the song down over a few bars.
• [Big Finish] — a climactic ending with held notes instead of a fade.
• [End] — a hard stop. This is the best fix for songs that mumble into a thirty-second fade-out.

A dependable skeleton looks like this:

If you would rather not hand-write the words for that skeleton, our AI Lyrics Generator produces original lyrics with the structure tags already in place, ready to paste straight into the lyrics box.

Style tags by category, with examples

Short comma-separated tags are easier to debug than flowing prose: when a generation goes wrong, you can remove one tag at a time and see what changes. The categories that reliably steer the model:

• Genre: synthwave, boom bap, melodic dubstep, outlaw country, bossa nova, pop punk. One or two; three is the ceiling.
• Mood: melancholic, triumphant, hazy, menacing, euphoric, bittersweet.
• Era: 80s, 90s golden era, Y2K pop, 1960s soul. Era tags pull in period-appropriate production on their own.
• Vocal type: female vocal, male baritone, airy falsetto, rap verse with sung chorus, gospel choir, or “instrumental, no vocals” if you want none.
• Instrumentation: analog synths, 808 bass, fingerpicked acoustic guitar, string quartet, brass stabs.
• Production: lo-fi, tape saturation, punchy drums, wide stereo, sparse arrangement, wall of sound.

A working formula: one or two genres, one mood, a vocal type, then two or three instrumentation or production details. That is usually six to nine tags, which is the sweet spot.

5 complete example prompts

1. Dreamy bedroom pop

Expect a mellow, slightly washed-out track where the chorus lands gently rather than exploding. Add “big soaring chorus” if you want more lift.

2. 90s boom bap

Expect swung drums and a sampled feel. Keep rap verses to eight or twelve lines so the flow fits the bar count.

3. Epic cinematic trailer (instrumental)

Leave the lyrics box empty or use only tags like [Instrumental Break] and [Big Finish]. Stating “no vocals” twice, as above, meaningfully cuts the chance of a stray choir appearing.

4. Modern country ballad

Expect a slow, spacious arrangement. The pedal steel tag does a lot of work here; without it you often get generic acoustic pop.

5. Dark synthwave

Expect a steady pulse and a vocal that sits behind the synths. Swap the vocal tag for “instrumental” and this doubles as a solid backing-track prompt.

Common mistakes that wreck generations

• Contradictory tags. “Intimate acoustic ballad, festival EDM drop” averages into something that is neither. Pick one energy level per song.
• Over-stuffing. Twelve genres do not make the song twelve times more interesting; they make it beige. Six to nine focused tags beat twenty-five every time.
• Artist and band names. Suno blocks or silently ignores “in the style of” plus a real artist. Describe the ingredients instead: rather than naming a 90s grunge band, ask for “90s grunge, raw male vocal, fuzzed-out guitars, quiet-loud dynamics.” You get closer to the sound and the prompt actually runs.
• Stage directions as plain text. “(solo here)” in the lyrics box tends to be sung. Use [Instrumental Break].
• No ending tag. Without [End] or [Big Finish], many songs trail off into repetitive fade-out mumbling.

Suno prompt cheat sheet

• Style box: 1–2 genres, 1 mood, vocal type, 2–3 sound details.
• Lyrics box: words to sing plus square-bracket structure tags.
• Reliable order: Intro → Verse → Pre-Chorus → Chorus → Verse → Chorus → Bridge → Chorus → Big Finish → End.
• Repeat chorus text verbatim under every chorus tag.
• Write “instrumental, no vocals” for vocal-free tracks.
• Never use real artist names; describe their sound instead.
• Generate 2–3 takes per prompt; the same prompt varies a lot between runs.
• Keep a note of prompts that worked. Future you will be grateful.

That is genuinely most of the craft. The rest is iteration: change one tag, regenerate, listen, repeat.