GPS Accuracy Explained
GPS accuracy by the numbers — ~4.9 m smartphone (95%), ~1-2 m with SBAS, ~1-2 cm with RTK, the full error budget (ionosphere, multipath, geometry), and what's tunable.
By Steve K.. Published . Last updated .
Civilian GPS achieves about 4.9 m horizontal accuracy at the 95th percentile under open sky per the 2020 SPS Performance Standard. Augmentation improves this — WAAS / SBAS reaches ~1-2 m, RTK reaches 1-2 cm, and carrier-phase post-processing reaches sub-centimetre.
GPS accuracy is not one number — it is a stack of error sources, each with its own magnitude and mitigation (the underlying mechanics live on /learn/how-gps-works). The headline "5 m" of a smartphone is the result of dozens of corrections applied to a raw satellite signal that, if uncorrected, would be tens of metres off. The WGS-84 datum (/learn/wgs84-explained) is the reference frame all this accuracy is measured against. This article runs the published accuracy commitments, the typical observed performance, the full error budget by source, and the augmentation paths that take consumer GPS from 5 m to 2 cm. The companion pillar /learn/how-gps-works covers the underlying mechanics; /learn/differential-gps and /learn/rtk-gps cover augmentation in depth.
The published commitments
GPS.gov publishes formal performance standards that the US Air Force guarantees. These are minimum commitments — observed performance is typically better.
| Metric | Commitment (95%) | Observed (95%, open sky) | Source |
|---|---|---|---|
| Horizontal positioning | ≤ 9.0 m | ~4.9 m | GPS.gov SPS PS 2020 §3.4 |
| Vertical positioning | ≤ 15.0 m | ~7.0 m | GPS.gov SPS PS 2020 §3.4 |
| Time transfer accuracy | ≤ 40 ns | ~10 ns | GPS.gov SPS PS 2020 §3.5 |
| User Range Error (URE) | ≤ 8.0 m | ~0.5-1 m | GPS.gov SPS PS 2020 §3.3 |
| Constellation availability | ≥ 99% (≥ 21 of 24 satellites) | ~100% typically | GPS.gov SPS PS 2020 §4 |
| PDOP ≤ 6 | ≥ 98% of time | ~99.9% | GPS.gov SPS PS 2020 §3.1 |
The 9.0 m and 15.0 m commitments are the formal numbers the US government guarantees for civilian use. The 4.9 m typical horizontal performance comes from FAA observed-data analysis (under open sky with good geometry) and is the de-facto baseline most receiver manufacturers quote.
Accuracy by receiver class
A consumer's actual accuracy depends mostly on what kind of receiver they have.
| Receiver class | Horizontal 95% | Vertical 95% | Cost | Example use |
|---|---|---|---|---|
| Smartphone (single-frequency, embedded) | 4-10 m | 8-20 m | Built into phone | Maps, ride-share, hiking apps |
| Smartphone (dual-frequency L1+L5) | 1-3 m | 3-6 m | Built into newer phones (iPhone 14+, Pixel 5+) | Improved navigation |
| Dedicated handheld (Garmin etrex) | 2-5 m | 4-8 m | $100-300 | Hiking, geocaching |
| SBAS-augmented receiver | 1-2 m | 2-4 m | $200-1,000 | Aviation, marine |
| Survey-grade with RTK | 1-2 cm | 2-3 cm | $5K-30K | Construction, cadastre, BIM |
| Survey-grade with PPP / post-processed | 5 mm – 2 cm | 1-5 cm | $10K-50K | Geodetic monumentation, plate-tectonics |
| Reference station (CORS, IGS) | Sub-centimetre at fiducial | Sub-centimetre | Permanent installation | NGS NSRS, ITRF realization |
The 1,000× spread (5 m smartphone to 5 mm survey post-processed) comes from three things: more sophisticated signal processing (carrier-phase tracking rather than code), better antennas (choke-ring designs that reject multipath), and access to differential corrections (RTK base station nearby, or carrier-phase post-processing against CORS reference data).
The error budget
Total positioning error is the quadrature sum of independent error sources scaled by Dilution of Precision (DOP):
σ_position ≈ DOP × sqrt(σ_iono² + σ_tropo² + σ_eph² + σ_clock² + σ_multipath² + σ_noise²)
| Error source | Single-freq (m) | Dual-freq (m) | RTK / DGPS (m) | Mitigation |
|---|---|---|---|---|
| Ionospheric delay | 5-10 (solar-cycle dependent) | 0.1-0.5 (cancelled by dual freq) | ~0 (cancelled by base station) | Dual-frequency receiver; SBAS |
| Tropospheric delay | 0.5-2 (model-corrected residual) | 0.5-2 (model-dependent) | ~0 (cancelled) | Saastamoinen model; PPP residual |
| Satellite ephemeris | 1-2 | 1-2 | ~0 (cancelled) | MCS broadcast corrections; precise post-eph |
| Satellite clock | 1-2 | 1-2 | ~0 (cancelled) | MCS broadcast corrections |
| Multipath | 0.5-1 open; 5-100 urban | Same | Same; partially cancelled | Choke-ring antenna; receiver algorithms |
| Receiver thermal noise | 0.1-1 | 0.1-1 | 0.001-0.01 (carrier phase) | Better receiver hardware |
| Selective Availability | 50-100 (1990-2000 only) | Same | N/A | Switched off 2000-05-01 |
The transition from single-frequency to dual-frequency cuts the ionospheric error from the dominant term (5-10 m) to negligible (0.1-0.5 m). This is why every modern smartphone (iPhone 14+, Pixel 5+) added L5 reception — it improved their horizontal accuracy from ~5 m to ~1-3 m without any other hardware change.
Dilution of precision (DOP)
The error budget above is multiplied by the geometric Dilution of Precision — how favourably the visible satellites are arranged around the receiver.
| DOP value | Quality | Geometric interpretation | Typical condition |
|---|---|---|---|
| 1.0 | Ideal | Satellites symmetrically distributed across the sky | Rare; possible in open sea with 8+ satellites |
| 1-2 | Excellent | Well-distributed; one near zenith | Open sky, multi-constellation receiver |
| 2-5 | Good | Typical open-sky GPS-only configuration | Most outdoor smartphone use |
| 5-10 | Marginal | Satellites clustered toward one side | Urban canyon partial obstruction |
| 10-20 | Poor | Very clustered geometry | Heavy obstruction; mountain valley |
| > 20 | Useless | Near-singular geometry | Tunnel mouth; not reliable |
DOP comes in variants: PDOP (position, 3D), HDOP (horizontal), VDOP (vertical), TDOP (time), GDOP (geometric, all 4 unknowns). Modern receivers report DOP in their status output. SPS PS 2020 guarantees PDOP ≤ 6 for at least 98% of the time globally.
Augmentation systems
Augmentation adds external corrections to the satellite signal, reducing one or more error sources.
| Augmentation | Provided by | Type | Accuracy gain | Coverage |
|---|---|---|---|---|
| WAAS | FAA (US) | SBAS — geostationary signal | ~5 m → 1-2 m | Continental US, Canada, Mexico |
| EGNOS | EUSPA (EU) | SBAS | ~5 m → 1-2 m | Europe |
| MSAS | Japan JCAB | SBAS | ~5 m → 1-2 m | Japan |
| GAGAN | AAI / ISRO | SBAS | ~5 m → 1-2 m | India + region |
| SDCM | Roscosmos | SBAS | ~5 m → 1-2 m | Russia + region |
| BDSBAS | CSNO (China) | SBAS | ~5 m → 1-2 m | China + region |
| RTK (Real-Time Kinematic) | Survey base station | Carrier-phase differential | ~5 m → 1-2 cm | ~20 km radius of base |
| DGPS (code differential) | Maritime beacons, NTRIP | Code-phase differential | ~5 m → 0.5-1 m | 100-1,000 km from base |
| PPP (Precise Point Positioning) | IGS, NRCan, OPUS | Post-processed precise ephemerides | ~5 m → 5 mm – 2 cm | Global (post-processed) |
The two key augmentation paths in modern use are SBAS (free, broadcast on geostationary satellites, automatic in most consumer aviation receivers) and RTK (purchased, requires base station within ~20 km). Both are now embedded in many high-end receivers — modern u-blox F9P boards combine multi-constellation, dual-frequency, and RTK in a ~$200 module.
Vertical accuracy is worse than horizontal
GPS vertical accuracy is typically 2-3× worse than horizontal. This is not a bug; it is a direct consequence of satellite geometry.
| Reason | Effect | Typical impact |
|---|---|---|
| No satellites below horizon | Half the sky unavailable; only upper hemisphere contributes | ~2× worse VDOP |
| Satellite distribution favours horizontal | GPS satellites span upper hemisphere; verticals constrain less | ~1.5-2× worse residual |
| Ellipsoidal vs orthometric confusion | Raw GPS gives ellipsoidal; users expect orthometric | Up to ~100 m perceived error |
| Atmospheric delay affects vertical more | Tropospheric delay maps almost directly to vertical | ~2× weighting in vertical |
A consumer smartphone with 5 m horizontal accuracy typically has 10-15 m vertical accuracy. Aviation applications fuse barometric altimetry with GPS altitude to get the better of both — pressure altimetry is excellent at relative altitude change, but absolute altitude needs a calibration value (QNH) that GPS-derived geometric altitude can provide.
Conditions that degrade accuracy
| Condition | Why accuracy degrades | Magnitude |
|---|---|---|
| Urban canyon (downtown high-rise) | Multipath from buildings, partial sky obstruction | 5-100 m horizontal possible |
| Dense forest canopy | Signal attenuation; partial obstruction; multipath | 5-20 m typical |
| Indoors | Signal attenuation through walls/roof; multipath dominant | Tens of metres if any fix at all |
| Near tunnels or under bridges | Brief signal loss; receiver coasts on Kalman filter | Drift proportional to time without signal |
| Solar maximum / geomagnetic storm | Ionospheric scintillation; phase noise | Single-freq: +5-15 m; dual-freq: +0.5-2 m |
| Polar latitudes (> 75°) | Few satellites overhead; poor geometry; high TEC | ~2-3× worse than mid-latitudes |
| Low-elevation satellites (< 10°) | Long signal path through troposphere; ground multipath | ~2-5× worse for the affected satellite |
Modern receivers exclude satellites below 5° or 10° elevation by default to mitigate the worst tropospheric and multipath errors. Multi-constellation receivers (GPS + Galileo + GLONASS + BeiDou) help with urban canyons by improving the chance that some satellite geometry is good enough.
The Selective Availability era
From 1990 to 2000, the US military intentionally degraded civilian GPS accuracy by adding a pseudo-random error to the broadcast signal — Selective Availability (SA).
| Era | SA status | Civilian accuracy (95%) | Trigger |
|---|---|---|---|
| 1990-2000 | On | ~50-100 m horizontal | Cold War |
| 2000-05-01 to present | Off | ~4.9 m horizontal (open sky) | Clinton executive order |
| 2007 onwards | New satellites built without SA capability | ~4.9 m (improving with modernization) | GPS III block |
On 2000-05-01 President Clinton ordered SA switched off, instantly improving civilian GPS accuracy by ~10×. The economic impact was estimated at tens of billions of dollars in the first few years, as applications previously not viable (precision agriculture, surveying, location-based services) suddenly became possible.
When better-than-5m accuracy is required
| Application | Required accuracy | Augmentation needed |
|---|---|---|
| Driving directions | 5-10 m | None (raw GPS adequate) |
| Lane-level autonomy | 0.5-1 m | SBAS + multi-constellation; or RTK |
| Precision agriculture (autosteer) | 5-30 cm | RTK |
| Construction layout, BIM | 1-2 cm | RTK |
| Cadastral surveying | 1-5 cm | RTK or PPK (post-processed kinematic) |
| Geodetic monumentation | < 1 cm | PPP, OPUS, or 24h+ static post-processing |
| Plate tectonics measurement | < 1 mm | Long-term carrier-phase post-processing |
Common misconceptions
Related
- How GPS Works— The pillar — the system this article reports the accuracy of
- Precision vs Accuracy in Coordinates— The general concept that informs the GPS-specific numbers
- Why GPS Is Not Always Accurate— Practical failure modes (when shipped)
- My Location tool— Reports browser-supplied GPS coordinates with accuracy estimate
- Methodology— How content is sourced and verified
Frequently asked questions
How accurate is GPS on a smartphone?
Per GPS.gov, a typical civilian smartphone or handheld GPS receiver achieves approximately 4.9 m horizontal accuracy under open sky, at the 95th percentile. With WAAS / SBAS augmentation (built into most modern smartphones), accuracy improves to roughly 1–2 m. Urban canyons, dense canopy, and indoor environments are significantly worse — 10–50 m is common in dense cities, and indoor positioning falls back to cellular triangulation or Wi-Fi (50–500 m). The 4.9 m figure assumes a clear sky view of multiple satellites and undegraded signals.
What is the GPS signal-in-space URE?
URE (User Range Error) is the error in the broadcast signal as it reaches a ground receiver, before the receiver's own errors are added in. Per the 2020 GPS Standard Positioning Service Performance Standard, the URE is specified as ≤2.0 m at the 95th percentile (often actually around 0.7 m for the modernized signal). URE is the satellite-side error budget; the receiver-side adds its own errors (multipath, atmospheric correction model accuracy, receiver noise), and the user-perceived horizontal accuracy is the geometric combination of all these factors.
How accurate is RTK GPS?
Real-Time Kinematic (RTK) GPS achieves 1–2 cm horizontal accuracy with a properly-set-up base station and dual-frequency receiver. RTK works by having a fixed reference station with known coordinates transmit correction data to a mobile receiver in real time — both receivers track the same satellites, so atmospheric and orbital errors largely cancel out. RTK is the standard for high-precision surveying, infrastructure monumentation, agriculture (precision planting), and increasingly autonomous vehicles. Setup is complex (base station within ~30 km, communications link), but the accuracy is order-of-magnitude better than uncorrected GPS.
What contributes most to GPS error?
For typical civilian GPS without augmentation, the major contributors are: ionospheric delay (~5 m if uncorrected; ~0.5 m with single-frequency models; ~10 cm with dual-frequency), satellite ephemeris and clock errors (~1 m), multipath (highly variable; 1–10 m in urban areas), tropospheric delay (~0.5 m), and receiver noise (~0.1 m). With SBAS corrections, ionospheric and ephemeris errors are largely removed; with RTK, all distance-correlated errors largely cancel. The dominant remaining error in modern systems is multipath, which is purely receiver-environment dependent.
How does GPS accuracy compare to other positioning systems?
Cellular triangulation: 50–500 m typical. Wi-Fi positioning: 5–50 m in well-mapped urban areas. Bluetooth beacon positioning: 1–5 m in venues with installed beacons. Inertial dead-reckoning: drifts at 10 m/min without correction, useful as GPS-outage bridging. Visual odometry (camera-based, used in autonomous vehicles): centimetre-level when GPS-fused. Survey-grade RTK GNSS: 1–2 cm. Survey-grade post-processed (PPP, no base station needed but slow convergence): 1–10 cm. GPS standalone falls in the middle of this range and is overwhelmingly the dominant outdoor positioning technology globally.
Why is GPS less accurate in cities?
Tall buildings cause two problems: signal blocking (you see fewer satellites, worsening geometry/DOP) and multipath (signals bounce off building faces, arriving via indirect paths that confuse the receiver). Urban canyon multipath can add 5-100 m of error to consumer GPS. Modern receivers use multiple GNSS constellations (GPS + Galileo + GLONASS + BeiDou) and Kalman filtering to mitigate, but accuracy degrades from ~5 m open-sky to 10-50 m in downtown environments.
Why does GPS sometimes show me in the wrong place?
Three common causes: (1) Cold start — receiver hasn't downloaded the satellite almanac yet and is using stale data; wait 30 seconds. (2) Multipath — buildings, trees, or terrain causing signal reflection; move to open sky. (3) Insufficient satellites — fewer than 4 visible, geometry is poor; receiver may report a "low accuracy" warning. iOS and Android both expose an accuracy radius in their location APIs; treat it as the 68% confidence circle, not a guarantee.
What is the accuracy of GPS altitude?
GPS altitude is typically 2-3× worse than horizontal because no satellites are visible below the horizon. A phone with 5 m horizontal accuracy might have 10-15 m vertical accuracy. Additionally, raw GPS altitude is ellipsoidal (above the WGS-84 ellipsoid); to get orthometric height ("above mean sea level") requires subtracting the geoid undulation from EGM2008 (global) or GEOID18 (US). Most consumer GPS apps apply this conversion silently.
Sources
- GPS.gov — GPS.gov — Accuracy and performance specifications · https://www.gps.gov/systems/gps/performance/accuracy/ · Accessed .
- GPS.gov — GPS.gov — Signal-in-Space User Range Error (URE) · https://www.gps.gov/technical/ps/2020-SPS-performance-standard.pdf · Accessed .
- FAA — FAA — WAAS performance and accuracy · https://www.faa.gov/about/office_org/headquarters_offices/ato/service_units/techops/navservices/gnss/waas · Accessed .
- NOAA NGS — NGS — CORS accuracy specifications · https://geodesy.noaa.gov/CORS/ · Accessed .
Cite this article
APA format:
Steve K. (2026). GPS Accuracy Explained. Coordinately. https://coordinately.org/learn/gps-accuracy-explained
BibTeX:
@misc{coordinately_gpsaccuracyexplained_2026,
author = {K., Steve},
title = {GPS Accuracy Explained},
year = {2026},
publisher = {Coordinately},
url = {https://coordinately.org/learn/gps-accuracy-explained},
note = {Accessed: 2026-06-05}
}