Get Quake Live dialed in.

Jump to what you need

Setup, testing and fixing the weird stuff

01 · Start here

Stop looking for the one correct config.

Open five strong players' configs and you will find five different resolutions, sensitivities, crosshairs and ideas about mouse acceleration. That is normal. The part worth copying is the habit: know what every unusual line does and keep enough notes to rebuild the setup later.

You will see three little tags as you go. Technical means there is something concrete to test. Preference means reasonable players disagree. Advanced means leave it alone until you have a specific problem.

02 · Five minutes well spent

Make a backup. Seriously.

Before experimenting, open the console and write a snapshot:

writeconfig backup-before-guide.cfg

For a smaller client-oriented export, current Quake Live also provides:

writeClientConfig backup-client.cfg

Then quit the game and copy the entire active account folder somewhere safe. In the Steam release on Windows, Bethesda documents the player-specific location as:[1]

…\Steam\steamapps\common\Quake Live\<SteamID64>\baseq3\

That folder contains generated configuration such as qzconfig.cfg and repconfig.cfg, your optional autoexec.cfg, demos and other per-player material. Steam libraries may live on another drive. Search beneath the installed Quake Live directory rather than assuming it is on C:.

03 · Check your work

The console tells you what the game actually accepted.

Enable the console in Settings, then use the grave/tilde key. On some layouts the legacy combination is Ctrl + Alt + `. Setting com_allowConsole 1 allows the single console key where supported.

04 · A clean starting point

Start here, then make it yours.

This block is short on purpose. Check your current values first, use the menus when they are easier, and delete any line you cannot explain.

// Smoothness and view
seta com_maxfps "250"       // use 125 if 250 is not stable in combat
seta cg_drawFPS "1"
seta cg_fov "105"
seta cg_zoomFov "40"

// Predictable Windows mouse baseline
seta in_mouse "2"           // raw input; apply with in_restart/relaunch
seta m_filter "0"
seta cl_mouseAccel "0"
seta cl_mouseAccelOffset "0"
seta cl_mouseSensCap "0"

// Readability
seta cg_simpleItems "1"
seta cg_speedometer "3"     // numeric horizontal velocity

// Current broadband baseline
seta rate "25000"
seta cl_timeNudge "0"

You will notice there is no packet voodoo here. Current Quake Live locks cl_maxpackets at 125, snaps 40 is not right for every server, and nobody can choose your sensitivity or VSync tolerance for you.

05 · Display

Get the picture clear first.

Start with native resolution

Your monitor's native resolution normally provides the clearest text, crosshair edges and distant silhouettes. Lower resolution can help weak hardware or create a preferred visual scale, but it is not an accuracy upgrade.

Use the Video menu first. Common console controls include r_fullscreen, r_mode, r_customWidth and r_customHeight. Because the current closed-source client has no complete official public cvar table, query each value in your own build before relying on a legacy mode-number chart.

// Common desktop/native request
seta r_fullscreen "1"
seta r_mode "-2"
vid_restart

// Custom mode when the menu cannot express it
seta r_mode "-1"
seta r_customWidth "2560"
seta r_customHeight "1440"
vid_restart

Fullscreen, VSync and refresh rate

Picmip is not a personality

r_picmip 0 is highest texture detail; larger values reduce it. Some old tournament configs turn maps into flat soup for gains that barely register on modern hardware. Try bloom, post-processing, dynamic lights, marks, brass, fast sky and simple items individually. Keep the ones that make the picture easier to read or the frame rate steadier.

06 · FPS and frame pacing

A steady 166 beats a messy 250.

cg_drawFPS 1 shows the frame counter. com_maxfps caps the client frame loop/render rate, with 250 the normal unmodified upper cap in current Quake Live; the server remains authoritative. Choose a cap the machine can hold while rockets, plasma, players, recording software and browser windows are all doing what they normally do.

The millisecond timing used by this engine means requested caps can land on effective steps such as roughly 100, 111, 125, 142, 166, 200 and 250 rather than every arbitrary integer. Confirm the observed result.

Advanced com_idleSleep 0 is sometimes used when a system will not hold its cap, but it can increase CPU use, heat and power. Treat it as a measured troubleshooting experiment—not baseline folklore.

07 · Field of view

Pick an FOV you can actually read.

cg_fov supports a broad range (historically documented as 10–130). Higher values show more peripheral arena space but make distant opponents appear smaller and increase edge distortion. Lower values enlarge the central view but sacrifice awareness.

Field-of-view starting points

Example	Tendency	Try it when…
90	Narrow / magnified	You value distant target scale and tolerate less peripheral information.
100–110	Balanced	You want a sensible starting band.
115–120+	Wide / fast-looking	You prioritize awareness and can read smaller distant targets.

seta cg_fov "105"
seta cg_zoomFov "40"
seta cg_zoomScaling "1"     // animated transition; 0 snaps
seta cg_zoomToggle "0"      // hold; 1 toggles
seta cg_zoomOutOnDeath "1"
bind MOUSE2 "+zoom"

cg_zoomSensitivity multiplies zoom sensitivity; zero invokes legacy Quake III-style scaling. Test rail flicks and tracking at both normal and zoom FOV. Changing FOV does not change physical centimeters per 360-degree turn, although perceived speed and target scale change dramatically.

08 · Mouse input

Write down the whole mouse setup.

Raw input on Windows

in_mouse 2 selects the Windows raw-input path and is the dependable baseline for most Windows players. Apply it with in_restart or a relaunch. Platform behavior differs, so Linux players should test rather than copy the Windows recommendation blindly.

seta in_mouse "2"
seta m_filter "0"
in_restart

m_filter 0 disables smoothing. Smoothing can make motion look gentler but may feel delayed; start without it.

Legacy sensitivity math

A naked sensitivity 2 tells another player almost nothing. With m_cpi 0, default m_yaw 0.022 and no acceleration:

At 800 DPI, sensitivity 2 and yaw 0.022, the result is approximately 25.98 cm/360. Measure a real full turn with tape; the formula assumes the inherited default yaw and linear input.

Quake Live natural units with m_cpi

Quake Live also supports a natural-unit mode. Set m_cpi to the mouse's calibrated effective CPI/DPI and keep m_yaw 0.022 (m_pitch 0.022 for matching vertical scale). Horizontal sensitivity then represents degrees per centimeter:[7]

Pick one system and stick with it. A sensitivity copied from somebody using m_cpi 0 becomes nonsense in a config using m_cpi 1600. Driver or firmware scaling can also make effective CPI differ from the number on the box. Leave m_yaw 0.022 and m_pitch 0.022 alone unless you know exactly why you are changing them; negative pitch inverts vertical look.

09 · Mouse acceleration

Acceleration is fine. Mystery acceleration isn't.

Linear input is easier to learn, reproduce and troubleshoot, so the neutral baseline is:

seta cl_mouseAccel "0"
seta cl_mouseAccelOffset "0"
seta cl_mouseSensCap "0"

Acceleration can still be a valid arena-FPS preference: slow hand motion can support fine tracking while fast motion produces a larger turn. Current Quake Live exposes cl_mouseAccel, cl_mouseAccelOffset, cl_mouseAccelPower and cl_mouseSensCap. Tune one parameter at a time and record all of them with DPI and base sensitivity.

10 · Crosshair and HUD

Your crosshair has one job.

A good crosshair survives bright sky, dark stone, orange explosions, green effects and a bright enemy without covering a tiny distant target.

seta cg_drawCrosshair "2"      // preview styles in the menu
seta cg_crosshairSize "24"
seta cg_crosshairBrightness "1"
seta cg_crosshairPulse "0"
seta cg_crosshairHealth "0"

Historic documentation lists built-ins 1–21, while the current interface/build may expose more. Use the menu preview rather than trusting a screenshot of an old list. Hit feedback is controlled by cg_crosshairHitStyle, cg_crosshairHitColor and cg_crosshairHitTime; verify ranges in your build.

Make the important numbers boringly obvious

You should be able to catch health, armor, ammo, time and score with a glance. In team modes, add team status and chat without burying those basics.

seta cg_weaponBar "1"
seta cg_drawFullWeaponBar "0"
seta cg_drawTeamOverlay "1"
seta cg_drawItemPickups "1"
seta cg_speedometer "3"       // numeric horizontal velocity

cg_drawGun 0 hides the view weapon, 1 shows it with sway, and 2 shows it without walking sway. This is a space/feedback preference; it does not alter weapon damage or accuracy.

11 · Visibility and comfort

Make players easy to find.

Consistent bright enemy models reduce recognition time. A common example is:

seta cg_forceEnemyModel "keel/bright"

Other available bright models can be equally valid. Use cg_enemyHeadColor, cg_enemyUpperColor and cg_enemyLowerColor with the documented 0xRRGGBBAA format, then test across pale sky, green slime, orange light and dark industrial rooms. Team modes need a clearly different forced team model/color too.

Damage and motion comfort

seta cg_bob "0"
seta cg_kickScale "0"
seta cg_screenDamage "0"
seta cg_screenDamage_Self "0"
seta cg_screenDamage_Team "0"

Use literal 0 to disable screen-damage overlays. Malformed “transparent” color strings copied from modified setups can produce broken solid overlays. Bob, kick, flashes and weapon presentation are also accessibility choices; reduce what causes nausea, distraction or discomfort.

12 · Sound

Turn up information, not just volume.

Footsteps, jumps, weapon fire, teleporters, jump pads and item pickups reveal position and intent outside the camera. Start with the menu, then query the exact cvars available in your build.

seta s_volume "0.8"
seta s_musicVolume "0"
seta s_announcerVolume "0.8"
seta s_voiceVolume "0.8"

Those numbers are examples. A value copied from somebody else's config cannot tell how loud your headphones are. Use the lowest level that keeps the cues clear; hearing damage is a rotten trade for one more item pickup.

13 · Network

No cvar can make Texas closer to Frankfurt.

The servers behind this site are in Spring, Texas. Players nearby should see a short, steady route; somebody across an ocean will not. That is geography, not a broken config.

The baseline is pleasantly boring:

seta rate "25000"
seta cl_timeNudge "0"
seta cg_lagometer "1"

rate applies after reconnect. Current Quake Live constrains it to the modern range (roughly 8000–25000). cl_maxpackets is locked at 125 in current Quake Live, so old 63-versus-125 optimization advice is obsolete.[10]

Read the lagometer

cg_lagometer 1 shows the packet graph; 2 adds a client ping estimate. Use it to distinguish steady latency from jitter, rejected snapshots and actual packet loss. A good ping number does not prove a clean connection.

Advanced cl_timeNudge is constrained in Quake Live (commonly 0 through −20). Negative values do not lower real ping; they trade interpolation margin for apparent immediacy and can look rough under jitter. Start at zero.

Advanced snaps requests snapshot frequency, but the server controls and caps what it sends. Old “always use 40” or “match every server's sv_fps” recipes are server/mod dependent. Leave it alone unless the operator documents a reason.

cl_packetdup sends redundant user commands for loss resilience at an upload cost. Query its current value; do not change it without demonstrated packet loss. Keep cl_nodelta 0 for normal delta compression.

14 · Key binds

Put weapons where your fingers can reach them.

If reaching Rocket Launcher makes you let go of a movement key, the bind is in the wrong place. Railgun, Lightning Gun and Rocket Launcher should all be reachable while you keep moving. Scrolling through weapons is too easy to overshoot under pressure.

bind W "+forward"
bind S "+back"
bind A "+moveleft"
bind D "+moveright"
bind SPACE "+moveup"
bind CTRL "+movedown"
bind MOUSE1 "+attack"
bind MOUSE2 "+zoom"
bind TAB "+scores"

Weapon numbers

bind Q "weapon 5"    // Rocket Launcher
bind E "weapon 6"    // Lightning Gun
bind R "weapon 7"    // Railgun
bind F "weapon 8"    // Plasma Gun
bind 3 "weapon 3"    // Shotgun

Availability depends on map and mode. Build around your hand: side mouse buttons, wheel directions and nearby letters are all valid if they do not cause accidental switching.

Chat and team communication

bind T "messagemode"
bind Y "messagemode2"
bind V "say_team ^3MEGA AVAILABLE"

Keep scripted calls short and truthful. Do not flood chat, automate deceptive timers or attach large text blocks to routine actions.

15 · Config files

Keep autoexec.cfg boring.

Create autoexec.cfg beside qzconfig.cfg in the active SteamID64 baseq3 folder. Turn on file extensions and make sure Windows did not quietly name it autoexec.cfg.txt. The best autoexec is usually short enough to read without making coffee:

// autoexec.cfg
exec zipnar-client.cfg
echo "^2AUTOEXEC LOADED"

Then put deliberate settings in zipnar-client.cfg:

// View
seta cg_fov "105"
seta cg_zoomFov "40"

// Input — 800 DPI, linear, ~26 cm/360
seta in_mouse "2"
seta m_cpi "0"
seta sensitivity "2"
seta m_yaw "0.022"
seta cl_mouseAccel "0"

// Readability
seta cg_simpleItems "1"
seta cg_drawFPS "1"
seta cg_forceEnemyModel "keel/bright"

// Weapons
bind Q "weapon 5"
bind E "weapon 6"
bind R "weapon 7"
bind F "weapon 8"

echo "^2ZIPNAR CLIENT CONFIG LOADED"

Why split files?

A larger setup can use exec video.cfg, exec mouse.cfg, exec binds.cfg and exec practice.cfg. Comments should explain every unusual value. If a setting resets, search for duplicate definitions, confirm the load message, query the value after joining, and check whether the server clamps it.

16 · Demos

Record the match before rewriting your config.

record practice-rail-01
stoprecord
demo practice-rail-01

Quake Live also supports automatic match actions through cg_autoAction: historically documented values are 0 none, 1 demo, 2 end screenshot, 3 both. Query the current value and test on a disposable match before trusting automation. Historically documented cl_demoRecordMessage modes are 0 hidden, 1 full recording text and size, and 2 a small REC indicator; verify the effective value in your console.[6][11]

Demos live under the player-specific baseq3\demos directory. Review one question at a time:

• Was the correct weapon ready before contact?
• Did the crosshair disappear against a background?
• Did tracking oscillate around the target?
• Did a turn run off the mousepad?
• Did FPS fall during real combat?
• Did the lagometer show loss or jitter?
• Was an audio cue audible and understood?
• Was the HUD readable without staring?

One ugly match proves almost nothing about your sensitivity. Look for the same failure showing up again before changing anything.

17 · Ready-made starting points

Four places to start.

These are small pieces to merge into your setup, not replacement configs. Read the comments and steal only what you understand.

seta com_maxfps "250"
seta r_swapInterval "0"
seta cg_fov "105"
seta cg_zoomFov "40"
seta in_mouse "2"
seta m_filter "0"
seta cl_mouseAccel "0"
seta cg_simpleItems "1"
seta cg_drawFPS "1"
seta cg_forceEnemyModel "keel/bright"

seta com_maxfps "125"
seta cg_fov "100"
seta in_mouse "2"
seta m_filter "0"
seta cl_mouseAccel "0"
seta cg_simpleItems "0"
seta cg_drawFPS "1"

seta cg_fov "100"
seta cg_zoomFov "45"
seta in_mouse "2"
seta cl_mouseAccel "0"
seta cg_simpleItems "1"
seta cg_speedometer "3"
bind Q "weapon 5"
bind E "weapon 6"
bind R "weapon 7"
bind F "weapon 8"
bind MOUSE2 "+zoom"

seta cg_fov "100"
seta cg_bob "0"
seta cg_kickScale "0"
seta cg_screenDamage "0"
seta cg_screenDamage_Self "0"
seta cg_screenDamage_Team "0"
seta cg_crosshairBrightness "1"
seta cg_crosshairPulse "0"

18 · Testing changes

Change one thing and go play.

1. Save the baseline. Record resolution, refresh, cap, FOV, zoom FOV, DPI, sensitivity, cm/360, acceleration, crosshair and model.
2. Name the problem. “Crosshair disappears against sky” is testable; “settings feel bad” is not.
3. Change one variable. Write the old and new values down.
4. Use a controlled drill. Same map, route, weapon and duration.
5. Test a real match. Empty-room smoothness is not combat stability.
6. Review evidence. Demo, FPS behavior, lagometer and physical comfort.
7. Keep or revert. If the change did not solve the named problem, restore the baseline.

19 · Internet archaeology

Old advice that refuses to die.

20 · When something breaks

Fix the smallest broken thing first.

21 · Quick answers

The questions people actually ask.

22 · Keep this nearby

Command reference.

Final configuration checklist

• Chosen resolution is intentional
• Actual refresh rate verified
• FPS stable during busy fights
• Raw/selected input path verified
• DPI, sensitivity and cm/360 recorded
• Acceleration intentionally on or off
• Crosshair survives bright and dark scenes
• Enemies and teammates distinguishable
• Health, armor, ammo and time readable
• RL, LG and RG directly bound
• Audio clear at safe volume
• Network tested for loss and jitter
• Autoexec prints its load marker
• Working folder backed up
• Changes tested one at a time

23 · Where this came from

Sources and image credits.

Quake Live's current executable is closed-source and does not publish a complete authoritative table of every client cvar. This guide therefore distinguishes current official Bethesda support, current-executable-facing technical documentation, Quake Live-specific historical developer material, and inherited Quake III source behavior. Historical defaults are labeled as historical; enter the cvar in your own console to see the effective value.

Technical sources

• [1] Bethesda Support — Quake Live configuration-file locations
• [2] Bethesda Support — recording and playing Quake Live demos
• [3] Bethesda Support — high ping and routing
• [4] Bethesda Support — resetting Quake Live configuration
• [5] Quake Live reverse-engineered/current command documentation
• [6] Quake Live advanced client options / preserved developer documentation
• [7] Quake Live m_cpi natural-units explanation
• [8] Quake Live Steam configuration guide (used cautiously for current/removed behavior)
• [9] id Software Quake III source — inherited yaw/input math context
• [10] Quake Live Steam discussion — locked cl_maxpackets, rate range and reconnect behavior
• [11] ESReality Quake Live forum — compact demo recording indicator

Image credits

• PC setup — Brian Wong, source, CC BY-SA 2.0; resized/WebP derivative.
• Gaming mouse — CAPTAIN MEDUSA, source, CC BY-SA 4.0; resized/WebP derivative.
• Gaming keyboard — YannikUllersperger, source, CC0; resized/WebP.
• Keyboard areas — Daniel Beardsmore, source, public domain.
• Gaming headset — Dinopkk, source, CC BY-SA 3.0; resized/WebP derivative.
• Ethernet cable — Dmitry G, source, CC BY-SA 3.0; resized/WebP derivative.
• Quake Live railgun gameplay — official id Software / Bethesda Softworks promotional screenshot, Steam source page; resized/WebP editorial derivative.
• Quake Live bright-model gameplay — official id Software / Bethesda Softworks promotional screenshot, Steam source page; resized/WebP editorial derivative.
• Aspect ratios — Indrek, source, public domain.

This is an unofficial community guide and is not affiliated with or endorsed by id Software, Bethesda Softworks, ZeniMax or Microsoft. Quake and related names are trademarks of their respective owners.