Your running shoe size is a fiction. At US men's 9, internal shoe length ranges from 255 mm to 281 mm across brands - a 26 mm spread that exceeds 3 full US sizes within one labeled size. FitMorph.4.4 Nike UK 11 measures 300 mm internally. Adidas UK 11 measures 284 mm. Same label, 16 mm difference - 1.5 UK sizes apart. "True to size" is metrologically meaningless without specifying a reference geometry.
This guide covers the actual dimensional physics of foot morphology, last engineering, sizing systems, and why the running shoe industry's approach to fit fails 37% of runners structurally. FitMorph.3.10 No affiliate links. No product rankings. Just the geometry.
The Truth Table: What You've Been Told vs. What's Actually Happening
| What people believe | What the physics shows | Why it matters | Source |
|---|---|---|---|
| Shoe size tells you if a shoe fits | A shoe size encodes 1-2 dimensions (length, sometimes width). The foot requires 6 dimensions for 93% shape coverage. 28-46% of morphological information is discarded at purchase. | You're selecting a 6D object using a 1D label. | FitMorph.1.8 |
| Feet swell during long runs | Foot volume increases +3% after 10 min running, then inverts at ~10 km. Marathon feet flatten (arch collapse), they don't swell. | Sizing up "for swelling" addresses the wrong mechanism. | FitMorph.2.4 |
| Width letters (B/D/2E) capture fit differences | At US men's 9, toe box width spans 67.6 mm (Adidas) to 90.4 mm (Altra) - a 22.8 mm spread exceeding 3 full width letters. Width grading in lasts exceeds actual biological variance. | Width letters are brand-specific, not standardized measurements. | FitMorph.3.6 FitMorph.3.3 |
| A single shoe width fits most people | A single width geometry accommodates at most 40.1% of the population within a given length class. Minimum 3 widths required for 90% coverage. | Most brands produce 1 width. Over half the population is structurally excluded. | FitMorph.1.4 |
| US/EU/UK size conversions are reliable | The ratio of barleycorn to Paris point (1.27) is irrational. Round-trip US-EU-US conversion can shift by up to 1 full US size. | Every conversion table is an approximation with unavoidable rounding errors. | FitMorph.4.2 |
| "Arch type" determines shoe category | Arch height distributes normally. No study has found trimodal or categorical distribution. Inter-rater reliability for categorical arch classification: near-random (kappa = 0.12-0.19). | Three arch types are artificial cutpoints imposed on a continuous spectrum. | FitMorph.6.2 |
| Breaking in shoes improves fit | The shoe is a manufacturing tool output, not an adaptive structure. Heat setting causes 1-3% polyester shrinkage. Lasting tolerances are plus/minus 4-5 mm = exactly one half-size. | Shoes don't adapt to your foot. Your foot adapts to the shoe (often pathologically). | FitMorph.3.7 |
The 8-Dimensional Foot: Why Shoe Sizes Fail
Your foot is not a rectangle
PCA on full 3D foot meshes requires exactly 6 principal components to explain 92.59% of shape variance. FitMorph.1.2 These map to: arch height (PC1), ball width + inter-toe distance (PC2), global foot width (PC3), hallux valgus angle (PC4), toe-length pattern (PC5), and midfoot width + toe angle coupling (PC6). Incorporating dynamic deformation raises the requirement to 8 PCs for ~95% variance.
Standard sizing encodes 1-2 of these dimensions. The information gap is 28-46 percentage points of morphological data discarded at the moment of purchase. FitMorph.1.8
The weak correlation problem
At a fixed foot length of 270 mm (North American males), the 5th-to-95th percentile spread is: forefoot width 15.8 mm, instep height 16.2 mm, heel width 12.7 mm. FitMorph.1.4 These dimensions exist on quasi-orthogonal axes. Holding length constant does not collapse their variance.
Arch height - the single largest independent shape variable - correlates with foot length at only R2 = 0.05-0.12. FitMorph.1.3 No commercial sizing system encodes arch height. The dimension that varies most between individuals is the dimension zero shoe manufacturers measure.
What the Brannock actually captures
The Brannock device measures 3 anatomical vectors with resolution of 4.23 mm per half-size (length), 5.59 mm per size (arch length), and 4.76 mm per letter (width). FitMorph.4.1 No peer-reviewed inter-rater reliability study of the device has been published despite near-universal market penetration since 1927.
Manual foot measurement tools show 95% limits of agreement of -18.8 mm to +12.7 mm - exceeding two full shoe sizes. FitMorph.4.1 The device is a 2D planimetric tool measuring a 3D dynamically deforming object. Parallax error of 3 mm is sufficient to cross a size boundary.
Dynamic Foot Deformation: What Happens Inside the Shoe
Your foot changes shape every stride
During running, the foot undergoes dramatic deformation that static measurements cannot predict:
- Forefoot width increases 6.0% at weight-bearing, reaching 9.7-10.4% under heavy loads. FitMorph.2.1 The metatarsal heads splay laterally as the longitudinal arch compresses.
- Foot length increases 2.2-5.6 mm during stance phase. In-shoe hallux-to-toebox gap narrows ~10 mm per stride. FitMorph.2.2
- Navicular displacement averages 5.6 mm during running (range: 0-20 mm, CV 45-55%). Static navicular drop does not predict dynamic arch behavior. FitMorph.2.3
The swelling myth
Foot volume increases +3% (~31 mL) after 10 min running, but at ~10 km, volume changes invert. FitMorph.2.4 The "feet swell two sizes" belief is a myth for typically hydrated marathoners. What actually happens is structural arch collapse under muscle fatigue - the foot flattens and spreads, becoming longer and wider despite stable or decreasing volume. FitMorph.2.5
Post-marathon, navicular height drops -4.8 mm immediately and continues to -6.6 mm at day 8 with no recovery. FitMorph.2.5 This is "flattening," not "swelling." The mechanism is intrinsic foot muscle fatigue causing arch collapse, not fluid accumulation.
The three-zone problem
The foot requires physically incompatible fit characteristics across its length. FitMorph.2.8 The heel zone requires high stiffness and low volume for calcaneal grip. The midfoot zone requires variable volume capacity (5-6 mm navicular vertical excursion). The forefoot zone requires high volume and high elasticity (~10% splay, ~5 mm elongation). These requirements demand directional compliance: expand transversely, resist laterally. No static structure accomplishes this.
Last Engineering: Why Every Shoe Is a Compromise
The last is not a foot model
A shoe last is a manufacturing tool parameterized by 15+ interdependent measurements: stick length (276 mm for men's US9D), ball girth (234 mm), ball width (~66% of girth = 156 mm), instep girth (~250 mm), heel width (58-63 mm), toe spring angle (15-25 degrees for running), and more. FitMorph.3.1
The Brannock measures width (2D linear projection). The last is engineered around girth (3D circumference). Conversion ratio: 1.33:1 - each width letter equals 6.35 mm girth but only 4.76 mm width. A consumer self-reporting "wide" has already lost a degree of freedom.
Grading errors compound at extremes
Last width grading exceeds actual foot-width growth by 3.5-5.9 mm per size step. FitMorph.3.3 Ball girth scales allometrically (exponent 0.743), but standard grading applies constant linear expansion. FitMorph.1.7 The industry practice of scaling all last dimensions proportionally with size is geometrically wrong - it over-scales width for large sizes and applies a height gradient with no biological basis.
At US12+: heel pocket too wide for calcaneus, causing heel slippage despite correct length (the "Bucket Effect"). At US6-7: upper doesn't scale down aggressively enough, creating boxy fit ("Small Size Compression"). FitMorph.3.3 A size 7 and size 13 are not geometrically similar objects - they are different shapes sharing a label. FitMorph.3.4
Curvature matters more than size
Lasts are classified by curvature: Linear (0-3 degrees, motion control), Semi-Curved (3-8 degrees, neutral trainers), Curved (8-15 degrees, racing). Shape mismatch greater than 5 degrees between foot axis and last curvature causes fit failure independent of correct length and width. FitMorph.3.5
A straight (rectus) foot type in a curved racing last (Nike Vaporfly, 12-15 degrees) experiences lateral toe crush and forced supination regardless of size accuracy.
Manufacturing tolerance exceeds sizing resolution
Heat setting causes 1-3% polyester shrinkage with plus/minus 5 degrees C oven tolerance. On size 10 (300 mm girth), 1% differential shrinkage = 3 mm volume variance - approximately half a size. Stick length tolerance is plus/minus 4-5 mm = exactly one half-size. FitMorph.3.7
CNC last milling achieves under 1 mm deviation, but finished shoes built on those lasts have multi-millimeter variances from stitching, materials, and assembly. Total tolerance stack-up means "size 10" from the same production run varies by 2-4 mm in internal length.
Width Systems: The 22.8 mm Lie
Width letters are brand-specific inventions
At men's US 9 medium, big-toe width spans 67.6 mm (Adidas Adizero Adios 7) to 90.4 mm (Altra Torin 7) - a 22.8 mm spread exceeding 3 full width letter steps. FitMorph.3.6 Internal shoe length at the same labeled UK size varies up to 16 mm between brands.
No enforceable international standard specifies last dimensions for a given size label. ISO/TS 19409 (last measurement standard) remains unpublished. FitMorph.3.6
The economics of width
A single model in one colorway with 15 lengths and 3 widths = 45 SKUs. With 10 colorways and 30 seasonal models = 13,500 SKUs. Tooling investment per model: over $85,000 before a single unit. FitMorph.4.6 Half-size increments in single width represent the economic optimum on the precision-versus-cost curve.
The result: a wide foot measuring true US 8 length sizes up to US 9.5 to acquire transverse girth, pushing the MTP joint out of alignment with flex grooves, misaligning arch support, and creating excess toe box void. FitMorph.4.6 FitMorph.4.7
The Unmeasured Dimensions: Instep, Toe Box, and Heel Shape
Instep height: the invisible variable
Three convergent lines of evidence establish statistical independence of instep height from foot length: R2 = 0.05-0.12 across studies with over 173,000 scans. FitMorph.5.3 At 270 mm male length, instep height spans 54.6-70.8 mm (16.2 mm range) - comparable to the entire width variation at the same length class, yet no sizing system encodes it.
Perceived stability correlates with peak dorsal pressure at r2 = 0.89-0.97. FitMorph.5.7 Runners perceive "fit" overwhelmingly through dorsal pressure - a parameter governed entirely by unmeasured dimensions. The dorsal foot surface has only 1-3 mm of skin separating upper material from bone.
Toe box volume
Toe box height and width are not strongly correlated (r = 0.57). FitMorph.5.4 A randomized crossover study (25 male marathoners) tested +8 mm vertical / +3 mm sagittal toe box expansion: hallux displacement reduced from 5.5 mm to 4.0 mm, toe deformation reduced 44%. Significant reductions in both vertical and anteroposterior GRF followed.
Morton's toe (2nd metatarsal longer than 1st) affects 10-35% of the population. FitMorph.5.5 Standard lasts taper toward a hallux-longest assumption, compressing the longer 2nd toe distally and causing subungual hematoma - the pathognomonic injury.
The sex-specific fit failure
After normalizing to foot length, 93% classification accuracy distinguishes male from female feet on absolute dimensions alone. FitMorph.5.6 Women have narrower heels relative to forefoot width, lower instep height, and greater ligamentous laxity. Width mismatches of 0-9 mm exist between female feet and men's running shoe lasts used for women's sizes.
The female foot presents an inverted triangle topology (narrow calcaneus, wide metatarsal heads). Linear downscaling of male lasts creates simultaneous opposing failures: excessive heel volume (calcaneal slippage) + insufficient forefoot volume (lateral metatarsal compression). FitMorph.5.6
Sizing Across Brands: The Conversion Problem
The US barleycorn (8.467 mm/size), EU Paris Point (6.667 mm/size), and Mondopoint (5-7.5 mm) use incommensurable base units. The ratio 8.467/6.667 = 1.27 is irrational, preventing clean size-for-size mapping. FitMorph.4.2
At US9 (~270 mm), US-EU length difference is ~2-3 mm. By US12, divergence reaches 5-6 mm (over 1 half-size). US system grades girth ~6.35 mm per size; EU system ~4.5 mm per size, explaining why EU-tooled brands (Salomon, Adidas) are perceived as "narrow" in the US market. FitMorph.3.2
Maximum discretization error: plus/minus 2.12 mm (US/UK half-sizes), plus/minus 3.33 mm (EU full sizes). Half-sizes at 4.23 mm sit precisely at the manufacturing tolerance threshold. FitMorph.4.3 Quarter sizes are metrologically meaningless: manufacturing tolerance of plus/minus 1-2 mm for assembled footwear means quarter-size distinction would be smaller than within-run variance.
3D scanning: precision without resolution
3D foot scanners achieve ICC 0.964-0.999 and accuracy deviation of only 1.6 mm (versus 14.7 mm for ink footprints). Volumental has scanned over 10 million feet. FitMorph.4.5 The scanner knows exact instep height. But if the retailer stocks only US 9 in D width, the 10-dimensional precision is funneled back into a 1-2 dimensional categorical label. The scanning technology has outpaced the sizing infrastructure.
The Arch Type Myth
Arch Height Index distributes normally with mean approximately 0.340 and SD 0.030. No study has demonstrated trimodal or categorical distribution. FitMorph.6.2 "Three arch types" result from imposing plus/minus 1 SD cutpoints, yielding ~68/16/16% - proportions expected from any normally distributed trait.
The MLA transitions between a Compliant Beam State (heel strike to midstance, absorbing energy) and a Rigid Truss State (midstance to toe-off, propulsive lever). FitMorph.6.1 Shoes rigidly supporting the arch throughout stance prevent beam-state shock absorption. Arch support should enable the transition, not prevent it.
Myths vs. Physics: 8 Fit Claims Tested
Myth 1: "Size up half a size for running shoes"
Physics: The standard 10-15 mm toe allowance is a static linear buffer that does not account for non-linear arch collapse, individual variance in arch flexibility, or rate of elongation. FitMorph.2.7 For high-arched feet with minimal dynamic elongation, half-size up creates excess volume. For flat-footed runners with 5+ mm dynamic elongation, half-size may be insufficient.
Myth 2: "Your foot type determines your shoe category"
Physics: Of 611 running shoe styles analyzed, only 63% fit as true US9 medium. The 37% failure concentrates in three structural clusters that have nothing to do with arch type: High-Instep/Wide-Ball, Narrow-Heel/Wide-Forefoot, and Straight-Foot Pes Planus. FitMorph.3.10 These are last engineering failures, not foot deficiencies.
Myth 3: "Wider shoes fix forefoot pain"
Physics: Width measures 1D linear distance between 1st and 5th metatarsal heads, ignoring transverse arch shape. A flat transverse arch may be wide linearly but very low in dorsal volume. A high pes cavus may be narrow linearly but require significant vertical volume. FitMorph.5.2 Forefoot pain requires volume assessment, not width alone.
Myth 4: "Rocker shoes fit the same as conventional shoes"
Physics: Standard toe spring is 15-20 mm; rocker shoes reach 30-50 mm. A Hoka with 40 mm toe spring reduces effective usable flat-bed length by ~20 mm versus a traditional shoe at same stick length. FitMorph.3.9 Hoka's Active Foot Frame seats the foot inside midsole sidewalls, creating additional width constraint. A Hoka "Wide" (2E) often matches internal volume of a standard Brooks "D."
Myth 5: "Carbon plate shoes fit like regular shoes"
Physics: Carbon-fiber plates reduce natural MTP dorsiflexion. If the shoe's flex point misaligns with the runner's MTP joints by over 3 mm, a levering effect lifts the heel during late stance, requiring tighter heel counter fit and exacerbating the conflict between forefoot splay and rearfoot lockdown. FitMorph.2.6
Myth 6: "Women's shoes are just smaller men's shoes"
Physics: Width mismatches of 0-9 mm exist between female feet and men's lasts used for women's sizes. The female foot has a fundamentally different topology - narrow heel relative to wide forefoot. FitMorph.5.6 Linear downscaling produces simultaneous heel slippage and forefoot compression.
Myth 7: "Your shoe size stays the same throughout life"
Physics: Ball width increases +3.1-4.0 mm per decade. Over a 30-year span (age 30 to 60), the forefoot may widen over 10 mm while foot length shows no significant change. FitMorph.5.8 Hallux valgus prevalence rises from ~23% to ~50% in women over 70. Your shoe size does stay the same. Your foot does not.
Myth 8: "3D foot scanning will solve the fit problem"
Physics: The scanning technology is extraordinarily precise (1.6 mm accuracy). FitMorph.4.5 The problem is that measurement must project onto a categorical system encoding only 1-2 dimensions. Brands maintain only 3-4 base last families with 3-4 widths = 12-15 discrete configurations - insufficient for an 8-dimensional shape space. FitMorph.3.10
What to Actually Look For When Fitting Running Shoes
1. Measure in the afternoon, after a run
Arch height drops ~10 mm from sitting to standing. Post-run, the foot flattens further with no recovery for days. FitMorph.1.6 FitMorph.2.5 Measure under conditions that approximate your worst-case scenario.
2. Fit by girth, not by width letter
Girth (3D circumference) captures more information than width (2D linear projection). FitMorph.3.1 A shoe that feels tight across the top of the foot may need a different last shape, not a wider size.
3. Know your heel-to-forefoot ratio
If your heel is narrow relative to your forefoot ("duck foot" pattern), no single standard last will serve both zones. FitMorph.3.10 Consider brands with independent heel/forefoot engineering or use aftermarket heel grip solutions.
4. Check MTP flex-point alignment
The shoe's flex groove should align with your MTP joint. Sizing up for width shifts this alignment. FitMorph.4.7 Misalignment causes biomechanical fatigue, heel slippage, altered gait, and accelerated material breakdown.
5. Test toe box volume, not just width
Toe box height matters independently of width (r = 0.57 correlation only). FitMorph.5.4 Press on the shoe top above each toe while standing. If any nail contacts the upper, the vertical volume is insufficient for your foot.
6. Ignore cross-brand size conversions
Your size in Nike tells you nothing about your size in Adidas. FitMorph.4.4 Start from foot length in mm and use each brand's specific size chart. Expect to own 2-3 different labeled sizes across your shoe rotation.
7. Account for last curvature
If you have straight feet (rectus foot type), avoid curved racing lasts (8-15 degrees). FitMorph.3.5 The curvature mismatch causes lateral toe crush that no amount of sizing adjustment fixes.
FAQ
Why do my running shoes fit differently from brand to brand?
No enforceable international standard specifies internal dimensions for a given size label. At US men's 9 across brands: internal length varies 26 mm, internal width varies 16.2 mm, and manufacturer-reported drop deviates from measured drop by -46% to +40%. FitMorph.4.4 Each brand's sizing is entirely self-referential.
Should I size up for marathon running?
Not because of swelling - that myth is not supported by evidence beyond 10 km. FitMorph.2.4 However, arch collapse under fatigue lengthens the foot structurally by several millimeters and continues for days post-race. FitMorph.2.5 If you have flexible arches, you may need extra toe room for the structural flattening that occurs in the final miles.
How can I tell if my shoes are the wrong shape (not just wrong size)?
Three clusters account for most structural fit failures: numbness on top of the foot (instep too shallow for your dorsum), heel slippage with adequate forefoot fit (narrow heel relative to forefoot width), and lateral pinky toe pain (curved last on a straight foot). FitMorph.3.10 All three indicate shape mismatch, not size error. Changing size will not resolve them.
Do I need wide shoes if I have bunions?
Hallux valgus (bunions) averages 8-16 degrees across populations, with 19% prevalence overall (23.74% in women). FitMorph.1.5 You need medial forefoot volume, not necessarily overall width. A shoe with a wide toe box but standard midfoot (Altra, Topo Athletic) addresses the geometry better than a uniformly wide last that creates excess heel volume.
Why do some shoes feel tight even in wide sizes?
Width measures only the linear distance between the 1st and 5th metatarsal heads. If your instep is high, the dorsal volume (unmeasured by any sizing system) is insufficient regardless of width grade. FitMorph.5.3 Dorsal foot surface has only 1-3 mm of tissue separating shoe material from bone. FitMorph.5.7 Small volume deficits create disproportionate pressure sensations.