How deep should I squat for strength: parallel vs ATG vs quarter

Squat to parallel as the default — hip crease at or below the top of the kneecap, the IPF/USAPL competition standard. ATG (below parallel, full bottom position) and quarter squats both exist for specific reasons, but neither replaces parallel for general strength training. The strength-standards multiplier ladder this site uses is calibrated to parallel; quarter squats let you load the bar heavier without producing the strength the multiplier ladder is measuring.

The “deep squats hurt knees” line is durable enough that it's worth meeting it head-on. A 2013 systematic review in Sports Medicine (Hartmann, Wirth, Klusemann; 164+ articles) drew the opposite conclusion: at training-realistic loads, deep squats are not higher-injury-risk than half or quarter squats — the combination that actually drives high-tissue-stress conditions is supramaximal load on shallow range of motion. This page walks through the joint mechanics (Schoenfeld 2010), what the systematic review actually concluded, and the narrow cases where ATG or quarter squats earn a place in a strength program. For where the depth assumption intersects with band reading, see squat standards by age and bodyweight.

What “parallel” actually means: hip crease below kneecap

The competition-legal definition is the cleanest one to anchor to: at the bottom of the rep, the surface of the legs at the hip joint must be lower than the top of the knees. That's the IPF/USAPL technical-rules standard, and it's the depth the strength- standards multiplier ladder used across this site is calibrated to. Three separate gym-floor proxies get loosely called “parallel” and only one of them clears the rule:

Source: IPF Technical Rules — squat depth criterion. Quarter / above parallel / thigh-flat are gym-floor descriptors; only hip-crease- below-kneecap is the published rule.

Take this seriously: thigh-parallel-to-floor and hip-crease-below- kneecap are not the same position. The thigh-flat cue gets the lifter most of the way there but typically leaves the hip a few degrees above the knee — a depth a meet judge would red-light. If your phone camera shows a flat thigh but a hip that's clearly above the top of the kneecap, you're a quarter-rep short of legal depth. That's the most common reason a lifter's gym-floor PR doesn't survive a competition platform.

Joint mechanics across depths

Schoenfeld's 2010 review in J Strength Cond Res, Squatting Kinematics and Kinetics and Their Application to Exercise Performance, lays out what happens at the ankle, knee, hip, and spine across the descent. The stated purpose is “to examine kinematics and kinetics of the dynamic squat with respect to the ankle, knee, hip and spinal joints and… to provide recommendations based on these biomechanical factors for optimizing exercise performance.” A short tour of what changes as depth increases:

• Knee. Tibiofemoral compressive force rises through the descent and peaks near maximum knee flexion. Anterior cruciate ligament (ACL) shear is highest at small knee-flexion angles (15–30°) and decreases as the knee moves into deep flexion; posterior cruciate (PCL) loading rises with depth and peaks near the bottom but is well-tolerated by a healthy PCL at training loads. The clinical implication is the opposite of the common-sense guess: shallow heavy reps load the ACL-relevant range most.
• Hip. Hip moment grows with depth — the deeper the position, the more the hip extensors (glute max, hamstrings, adductor magnus) drive the lift. This is a chunk of why the deep squat is one of the most effective lower-body exercises in the field: the hip is where the largest muscles act, and depth puts them on stretch.
• Spine.Spinal compressive load scales with bar weight, not depth. The depth-driven concern at the spine is posterior pelvic tilt at the bottom (“butt wink”) and the loaded lumbar flexion that follows; this is a technique problem, not a depth problem, and is fixed by training the anatomically available depth rather than forcing a deeper one than the lifter's ankle/hip mobility allows.
• Ankle. Ankle dorsiflexion is the gating mobility for upright deep squats; lifters with restricted dorsiflexion either heel-elevate (Olympic shoes / a small plate) or shift to a more hip-dominant low-bar pattern.

The framing that earns its place: depth changes the mix of which joints get loaded most, not the safety of the lift. The high-tissue-stress conditions are loaded shallow flexion (small knee-angle, big load) and spinal flexion under bar weight — neither of which is the same condition as “a deep squat with submaximal load and clean technique.”

The “deep squats hurt knees” myth, sourced

The cleanest published verdict on this is Hartmann, Wirth, and Klusemann, Analysis of the Load on the Knee Joint and Vertebral Column with Changes in Squatting Depth and Weight Load (Sports Medicine43(10):993–1008, October 2013). It's a systematic review covering 164+ articles on squat depth, knee-joint load, and spinal-column load. Two findings from the abstract that land directly on the “deep squats are dangerous” line:

With the same load configuration as in the deep squat, half and quarter squat training with comparatively supra-maximal loads will favour degenerative changes in the knee joints and spinal joints.

Provided that technique is learned accurately under expert supervision… the deep squat presents an effective training exercise for protection against injuries and strengthening of the lower extremity.

The mechanism the review walks through: shorter range of motion lets the lifter put more weight on the bar, and the spinal-column and knee-joint compressive forces in supramaximal half- or quarter-squat work climb past the loads a deep squat at training-realistic weights ever produces. The combination that actually accumulates joint micro-trauma is heavy load on shallow range of motion — which is structurally the quarter-squat training pattern, not the deep-squat pattern. Most lifters running a half- or quarter-squat program for “heavy lower-body work without going deep” have inverted the risk profile they were trying to dodge.

The directional take here matters: if your reason to avoid deep squats is “heavy deep squats wreck knees over time,” the published evidence runs the other direction at training-realistic loads. If your reason is anatomical (mobility limit, prior injury, rehab context, post-surgical instructions), that's a real conversation with a sports-medicine physician or a qualified strength coach — not a programming default. The systematic review also notes that the conclusion holds “provided technique is learned accurately under expert supervision”; the safety case is for trained, technique-clean deep squats, not random untrained loaded deep squats.

When ATG actually wins

ATG isn't more virtuous than parallel — it's a different stimulus with a different cost structure, and the use cases where it earns the depth are narrower than internet-strength culture suggests. Three real ones:

• Olympic weightlifting.The catch positions for the snatch and clean live below parallel by definition. A snatch caught at parallel is a missed lift; the bottom of an overhead squat is ATG by sport rule. If you're training the Olympic lifts, the full-bottom position is the lift, not a deeper variant of one.
• Mobility carry-over.Loaded full-range work in any joint tends to keep the joint's end-range usable; sedentary adults who can't reach a clean ATG position bodyweight have a mobility problem worth training around. Goblet squats to a deeper depth at light load are a useful entry point — the squat variants page covers the ratio.
• Sport-specific demand.Combat-sport, gymnastics, and some field-sport athletes benefit from competence in deep positions under load that a parallel-only training program doesn't build. This is not a strength argument; it's a specificity argument.

What ATG isn't a good case for: maximizing absolute squat numbers, hitting an intermediate-or-advanced multiplier band, or general lower-body hypertrophy. The same lifter taking a clean ATG rep will hold less load than they will at parallel — partly because of the longer time-under-tension at the bottom-out, partly because the leverages get progressively worse below parallel. If your reason to go ATG is “deeper is better,” you're paying load for a stimulus difference that doesn't buy more general strength. Train ATG when you have a sport reason; otherwise default parallel.

When quarter squats actually win

Quarter squats have real, narrow uses — but calling a quarter squat “a squat” is what causes most of the trouble around them. They're a different exercise, and treating them like a depth variant of the parallel squat is what produces the Hartmann 2013 risk pattern. Three honest use cases:

• Sticking-point partials. A lifter whose squat stalls in the top third of the lift can use paused-static partials from a power-rack pin to overload the position they fail at, usually at sub-1RM loads with strict tempo. This is rehab-adjacent programming, not a primary depth.
• Geared / equipped powerlifting lockouts. Equipment-supported lifters work supramaximal partials to acclimate to the loads gear lets them handle out of the bottom — a small slice of the powerlifting population, but a real one.
• In-season athlete maintenance. A field-sport athlete in-season may need lower-body load with minimal residual fatigue; high-box (above-parallel) squats produce a maintenance stimulus without the next-day soreness of a full-ROM session. Programming context, not a strength claim.

The take that earns its place: quarter squats are not “cheating” — they're a different exercise. The problem is that calling a 200 kg quarter squat “my squat” tells you nothing about the strength a parallel-or-deeper rep would measure, and the strength-standards multiplier ladder is calibrated to the parallel rep. If your only squat number comes from above-parallel depth, expect to land at least one band lower if you re-test to depth — that's why the squat-standards page spends a full section on the depth assumption.

How depth interacts with the strength-standards ladder

The bench multiplier ladder doesn't care about depth — a bench either touched the chest or it didn't. The squat ladder does, and ignoring it is the most common reason a band assignment reads too high. The multipliers used across this site (untrained 0.8× → novice 1.25× → intermediate 1.75× → advanced 2.5× → elite 3.0× of bodyweight for adult males; full ladder + female anchors here) assume a competition-legal squat — hip crease at or below the top of the kneecap.

A worked example to make the gap concrete. A 90 kg adult male squatting 200 kg quarter to a high-box reads as advanced on the ladder (200 ÷ 90 = 2.22×). The same lifter taking the same bar to legal depth might be moving 150–170 kg, which lands in the back half of intermediate (1.67–1.89×). Both reps are real; only one of them describes the strength the multiplier ladder was calibrated to. To read the band honestly, the depth underneath the load has to be honest first. If your number comes from above-parallel depth, expect to slide at least one band lower when you re-test to legal depth.

The full sourcing posture for this site — what counts as Tier 1 peer-reviewed numerical anchor versus training-population synthesis versus StrengthMath methodology — lives on the methodology page. The depth threshold itself is a published federation rule, not a calculator output; the multiplier ladder it interacts with is ExRx-aligned StrengthMath methodology, framed and rounded by this engine for in-house consistency.

Common questions

What counts as a 'parallel' squat?

Parallel as defined in the IPF/USAPL technical rules: hip crease at or below the top of the kneecap at the bottom of the rep. That's the competition-legal threshold and the depth the StrengthMath strength-standards multiplier ladder is calibrated to. Thigh-parallel-to-floor is a common gym-floor proxy but it's a slightly shallower position than hip-crease-below-knee-cap; if your phone-camera replay shows the thigh flat but the hip clearly above the knee, you're a few degrees short of competition depth.

Are deep squats actually bad for the knees?

No, not at submaximal training loads. Hartmann, Wirth, and Klusemann's 2013 systematic review in Sports Medicine (164+ articles) found that deep squats with submaximal loads do not contribute increased risk of injury to passive tissues, and that — at the same external load — half-squat and quarter-squat training with comparatively supramaximal weights actually favor degenerative changes in the knee joint and spinal column. The myth has the risk pattern backwards: load-on-shallow-ROM is where the high-tissue-stress condition lives, not depth-on-submax.

Should I train ATG (ass-to-grass) instead of parallel?

Only if you have a specific reason — Olympic-weightlifting catch position, sport-specific full-ROM demand, mobility carryover, or a competition that requires it. ATG is not more virtuous than parallel; it's a different stimulus with different costs (loaded spinal flexion at the bottom for some lifters, longer time-under-tension at the bottom-out, smaller absolute loads). For general strength, parallel-or-just-below is the higher-leverage default. The strength-standards multiplier ladder on this site assumes parallel.

Do quarter squats have any place in a strength program?

Yes, as a specialty tool — paused-static partials at a sticking point, lockout work for geared powerlifting, in-season maintenance for athletes who can't afford fatigue from full-ROM volume. They are not a substitute for a parallel squat. Calling a heavy quarter squat 'a squat' is what causes the band-assignment problem covered on the squat-standards page: a 200 kg quarter squat reads as advanced on a multiplier ladder calibrated to parallel, but the lifter can't reproduce that load to depth. Quarter squats are not cheating — they're a different exercise.

What about box squats and high-bar vs low-bar — do they change the depth answer?

Box squats and bar position are separate variables from depth. A box squat to a parallel-height box is still a parallel squat; a box set above competition depth is a quarter or high-box squat by another name. High-bar vs low-bar shifts torso angle and load distribution but doesn't change the depth definition. The variant question (high-bar / low-bar / front / paused / box / goblet) gets its own page; the depth question is independent.

Where to next

Once depth is settled, the natural next decision is bar position and variant. High-bar vs low-bar shifts torso angle and load distribution; front, paused, box, and goblet variants each carry a typical ratio against the back squat. The per-variant table and decision rule live at squat variants strength differentials. To convert a submax-rep set into an estimated 1RM before reading the band, run it through the 1RM calculator — the per-lift formula choices are covered at best 1RM formula. And to read your squat against the multiplier ladder this depth standard underpins, head back to squat standards by age and bodyweight. The broader calculator hub sits at the StrengthMath calculator hub.