A developer reference for recognizing, parsing and validating shipping and postal tracking numbers. Formats, tested regular expressions, and working check-digit algorithms for the major carriers and the global postal standard.
Track a parcel on InstantParcelsEvery courier encodes its tracking numbers differently. Some follow a global postal standard, some use a proprietary layout with a built-in check digit, and some are just a run of digits with no structure you can rely on. If you have ever tried to answer the deceptively simple question "which carrier does this tracking number belong to?", you already know there is no single, authoritative, machine-readable table for it. This page is an attempt at one.
It documents the formats used by the major global carriers and national postal operators, gives you a tested regular expression for each, and includes working check-digit algorithms for the formats that have them. Use it to build a carrier detector, to validate user input before you hit a tracking API, or as a reference when a tracking number lands on your desk and you have no idea who to ask.
If you would rather skip the parsing entirely and just track a parcel across any of these carriers in one place, that is exactly what InstantParcels does: it aggregates tracking for USPS, UPS, FedEx, DHL and 600+ carriers and national postal operators behind a single lookup. This document is the open reference that sits underneath that kind of problem.
Two properties of a tracking number matter when you are trying to identify it programmatically: its shape and its check digit.
The shape is the pattern of letters and digits: how many characters, where the letters sit, and whether it ends in a two-letter country code. Shape is what a regular expression captures, and it is usually enough to narrow a number down to a small set of candidate carriers. The problem is that shapes overlap. A bare string of twelve digits could be a FedEx number, a Deutsche Post number, or something else entirely, because a plain numeric run carries no identifying marks.
That is where the check digit comes in. Several formats reserve one position for a value computed from the others, so that a typo or a random string can be rejected without ever contacting the carrier. When a format has a check digit, validating it turns a maybe into a near-certainty. When it does not, shape is all you have, and you should treat detection as a best guess rather than a fact.
Throughout this document, regular expressions assume the input has been uppercased and stripped of spaces and hyphens first. Carriers and shipping labels love to print tracking numbers with spaces (9400 1118 9922 ...) or in lowercase, and normalizing before you match saves you a lot of pattern noise.
normalized = raw.upper().replace(" ", "").replace("-", "")| Carrier | Typical shape | Example | Regex (normalized) | Check digit |
|---|---|---|---|---|
| USPS | 22 digits (IMpb) | 9400111899223197428497 | ^(94|93|92|95)\d{20}$ | MOD 10 |
| UPS | 1Z + 16 chars | 1Z999AA10123456784 | ^1Z[0-9A-Z]{16}$ | UPS 1Z |
| FedEx | 12 / 15 / 20 digits | 020207021381215 | ^\d{12}$ | ^\d{15}$ | varies |
| DHL Express | 10 digits | 1234567890 | ^\d{10}$ | none |
| DHL eCommerce | GM + digits | GM6000000000000000 | ^GM\d{16,20}$ | none |
| Royal Mail | S10, …GB | AB123456785GB | ^[A-Z]{2}\d{9}GB$ | S10 |
| Canada Post | 16 digits | 1234567890123456 | ^\d{16}$ | none |
| Australia Post | S10 or 10-12 digits | AB123456785AU | ^[A-Z]{2}\d{9}AU$ | S10 |
| China Post / EMS | S10, …CN | RB123456785CN | ^[A-Z]{2}\d{9}CN$ | S10 |
| India Post | S10, …IN | EE123456785IN | ^[A-Z]{2}\d{9}IN$ | S10 |
| Deutsche Post / DHL DE | 12 / 16 / 20 digits | RR123456785DE | ^\d{12,20}$ | none / S10 |
| La Poste / Colissimo | 13 chars | 6A12345678901 | ^[0-9A-Z]{2}\d{9}[0-9A-Z]{2}$ | none |
| PostNL | 3S… or S10 | 3SABCD1234567 | ^3S[A-Z0-9]{9,11}$ | none / S10 |
| DPD | 14 digits | 01234567890123 | ^\d{14}$ | none |
| GLS | 11-14 digits | 12345678901 | ^\d{11,14}$ | none |
| Evri (Hermes) | 16 digits | 1234567890123456 | ^\d{16}$ | none |
| International (any) | S10 | RB123456785DE | ^[A-Z]{2}\d{9}[A-Z]{2}$ | S10 |
The examples above are structurally valid and, where a check digit exists, they pass it. They are constructed for documentation and do not correspond to real shipments.
By far the most useful format to understand is S10, the Universal Postal Union standard for identifying international postal items. It is used by essentially every national postal operator for registered mail, EMS, insured parcels, and tracked cross-border items, which means a single pattern covers a huge share of the world's tracked mail.
An S10 identifier is always thirteen characters, in three parts:
R B 1 2 3 4 5 6 7 8 5 D E
|_| |_______| | |_|
| | | |___ ISO country code (2 letters)
| | |_______ check digit (1 digit)
| |_____________ serial number (8 digits)
|_____________________ service indicator (2 letters)The service indicator is two letters. The first letter denotes the class of service (for example E for EMS, R for registered, C for insured parcels, L for tracked letters, U for unregistered), and the second letter is assigned by the originating operator. The serial number is eight digits assigned by that operator. The check digit is computed from the serial. The final country code is the ISO two-letter code of the country that originated the item, which is what lets you route …US to the US, …DE to Germany, …CN to China, and so on.
A single regex matches all of them:
^[A-Z]{2}\d{9}[A-Z]{2}$Note that the nine digits in the middle are the eight-digit serial plus the one check digit.
The check digit is a weighted modulo-11 calculation over the eight serial digits. Weights are applied left to right:
weights = 8, 6, 4, 2, 3, 5, 9, 7Multiply each serial digit by its weight, sum the products, then:
C = 11 - (sum mod 11)
if C == 10 -> C = 0
if C == 11 -> C = 5Worked example using serial 47312482 (from the UPU specification):
4x8 + 7x6 + 3x4 + 1x2 + 2x3 + 4x5 + 8x9 + 2x7 = 200
C = 11 - (200 mod 11) = 11 - 2 = 9So RA473124829US is a well-formed S10 number: service RA, serial 47312482, check digit 9, origin US. The two special cases (result 10 becomes 0, result 11 becomes 5) exist so the check digit is always a single decimal digit.
Because the country code is literally embedded in the number, S10 is the one format where you can identify the originating country with certainty even when you cannot pin down the exact operator, since several operators in one country can share the same …XX suffix.
The United States Postal Service uses the Intelligent Mail package barcode (IMpb), which encodes into a tracking number that is typically 22 digits long, though 20-digit and older variants appear in the wild. Domestic USPS numbers usually begin with a recognizable prefix (92, 93, 94, or 95), and numbers printed from certain label systems carry a leading 420 followed by the destination ZIP code, which you strip before reading the tracking portion.
^(94|93|92|95)\d{20}$ # common 22-digit IMpb
^\d{20}$ # 20-digit variant
^420\d{5}(94|93|92|95)\d{20}$ # with leading ZIP routing blockUSPS also carries international inbound items under the S10 standard, so a number ending in US and matching the S10 pattern is very likely a USPS-handled item. The IMpb tracking number ends in a MOD 10 check digit, the same scheme used by GS1 barcodes. Working from the rightmost digit of the number excluding the check digit, weight digits alternately by 3 and 1, sum them, and the check digit is whatever brings the total up to the next multiple of ten.
For a deeper walk-through of how to read a USPS number, decode its status messages, and spot common tracking scams, InstantParcels has a full guide at Master US Post Tracking Numbers and a live lookup at instantparcels.com/couriers/usps.
UPS tracking numbers are the easiest of all to recognize: they begin with 1Z and are exactly eighteen characters long.
^1Z[0-9A-Z]{16}$The structure is 1Z, a six-character shipper account number, a two-digit service level, a seven-digit package identifier, and a final check digit. UPS also issues a handful of other formats, such as a nine-digit "InfoNotice"-style number, numbers beginning with T followed by ten digits, and long Mail Innovations numbers that hand off to USPS for final delivery, but the 1Z form is what you will see the overwhelming majority of the time.
The 1Z number carries its own check digit over the sixteen characters that follow the 1Z prefix (fifteen significant characters plus the check digit itself). Letters are converted to digits on a repeating cycle (A=2, B=3, up to H=9, then I=0, J=1, and so on), which is the same as (ASCII_code minus 63) mod 10 for an uppercase letter. Then the fifteen significant characters are summed with odd positions weighted once and even positions weighted twice, and the check digit brings the total to the next multiple of ten:
value(c) = c if digit else (ord(c) - 63) mod 10
total = sum of odd-position values + 2 x sum of even-position values
check = (10 - total mod 10) mod 10The canonical UPS example 1Z999AA10123456784 passes this check, which makes it a convenient test vector when you implement the algorithm yourself. Track UPS at instantparcels.com/couriers/ups.
FedEx is the format most likely to trip up a naive detector, because its numbers are plain digits with no letters and no reliable prefix, and they come in several lengths: 12 digits for many ground and express shipments, 15 digits for FedEx Ground, and 20 or 22 digits for numbers derived from a longer barcode. Some FedEx SmartPost numbers begin with 96 and run 20 or more digits, which overlaps with other carriers' ranges.
^\d{12}$
^\d{15}$
^\d{20}$
^\d{22}$Because a 12-digit run is not unique to FedEx, treat a bare numeric match as a candidate rather than a confirmed identification. When a number could plausibly be FedEx or a postal operator, the honest answer is to present both and let the tracking lookup resolve it, which is precisely the kind of ambiguity a multi-carrier tracker is built to absorb. Track FedEx at instantparcels.com/couriers/fedex.
"DHL" is really several different products with different number formats, so it pays to distinguish them.
DHL Express, the international air courier arm, uses a 10-digit air waybill number (occasionally 11). It is purely numeric with no check digit you can rely on externally.
^\d{10}$DHL eCommerce and DHL Parcel, the lower-cost e-commerce and domestic products, use longer numbers that frequently begin with GM, or with LX/RX on S10-style international pieces.
^GM\d{16,20}$
^(LX|RX)\d{9}[A-Z]{2}$Deutsche Post / DHL Paket in Germany issues 12-, 16-, or 20-digit domestic numbers, and handles inbound international mail under S10 with a …DE suffix. Because the DHL family spans air freight, e-commerce, and a national post all at once, a DHL match should always record which DHL product you think it is. Track the DHL family at instantparcels.com/couriers/dhl.
Most national postal operators do not have a bespoke tracked-item format of their own for international mail; they use S10, and you tell them apart by the country-code suffix. This is the single most powerful idea in carrier detection, because one pattern plus a two-letter lookup covers dozens of operators.
| Suffix | Operator |
|---|---|
GB | Royal Mail / Parcelforce |
CA | Canada Post |
AU | Australia Post |
CN | China Post / China EMS |
IN | India Post |
JP | Japan Post |
ES | Correos |
IT | Poste Italiane |
FR | La Poste |
NL | PostNL |
DE | Deutsche Post |
SG | Singapore Post |
BR | Correios |
For any of these, the pattern is the same ^[A-Z]{2}\d{9}XX$ with XX swapped for the suffix, and the S10 check digit described above applies. That means for the entire postal family you get free validation: if the MOD 11 check digit passes, you can be confident the number is at least well-formed and correctly transcribed, even before you query anything. Browse all supported operators at instantparcels.com/couriers.
Note that many operators also run purely domestic formats that are not S10, such as Canada Post's 16-digit domestic number and Australia Post's domestic article numbers, so the S10 suffix identifies international items specifically.
Beyond the national posts, Europe has a dense field of private parcel carriers whose formats are mostly plain numeric and therefore mostly ambiguous:
6A.3S + 9 to 11 characters) alongside S10 for international.For these, expect shape-based detection to produce sets of candidates rather than a single answer. A 16-digit number in a European context could be Evri or Canada Post; without a check digit there is no way to be sure from the number alone. This is not a flaw in your code; it is a genuine property of the numbering schemes, and the correct engineering response is to carry a candidate list forward until a tracking query disambiguates it.
Putting it together, a robust detector runs in three stages: normalize, match by shape in priority order, then confirm with a check digit where one exists.
Order matters. Evaluate the specific, unambiguous patterns first (a 1Z… is unmistakably UPS; an S10 suffix pins the origin country) and fall back to generic numeric patterns last. If you test ^\d{12}$ before you test the structured formats, you will misclassify numbers that merely happen to be twelve digits long.
import re
PATTERNS = [
("ups", r"^1Z[0-9A-Z]{16}$"),
("usps", r"^(94|93|92|95)\d{20}$"),
("royal-mail", r"^[A-Z]{2}\d{9}GB$"),
("china-post", r"^[A-Z]{2}\d{9}CN$"),
("dhl-ecommerce", r"^GM\d{16,20}$"),
("postnl", r"^3S[A-Z0-9]{9,11}$"),
("s10", r"^[A-Z]{2}\d{9}[A-Z]{2}$"), # any international postal item
("dpd", r"^\d{14}$"),
("fedex", r"^\d{12}$|^\d{15}$"),
("dhl-express", r"^\d{10}$"),
# ... extend as needed
]
def candidates(raw):
s = raw.upper().replace(" ", "").replace("-", "")
hits = [name for name, pat in PATTERNS if re.match(pat, s)]
return hits or ["unknown"]The function returns a list on purpose. For structured formats it will usually be a single entry; for bare numeric formats it may return several, and that is the truthful output. If you want a single best guess, prefer the match with a passing check digit, then the most specific pattern.
Three of the formats here have check digits worth implementing. All three are short, and all three let you reject a mistyped number instantly, before you spend an API call on it.
import re
def s10_valid(tn):
m = re.match(r"^[A-Z]{2}(\d{8})(\d)[A-Z]{2}$", tn)
if not m:
return False
serial, chk = m.group(1), int(m.group(2))
weights = [8, 6, 4, 2, 3, 5, 9, 7]
total = sum(int(d) * w for d, w in zip(serial, weights))
c = 11 - (total % 11)
c = 0 if c == 10 else 5 if c == 11 else c
return c == chk
def usps_valid(tn):
if not tn.isdigit():
return False
body, chk = tn[:-1], int(tn[-1])
total = sum(int(d) * (3 if i % 2 == 0 else 1)
for i, d in enumerate(reversed(body)))
return (10 - total % 10) % 10 == chk
def ups_valid(tn):
m = re.match(r"^1Z([0-9A-Z]{15})([0-9A-Z])$", tn)
if not m:
return False
body, chk = m.group(1), m.group(2)
val = lambda c: int(c) if c.isdigit() else (ord(c) - 63) % 10
total = (sum(val(c) for i, c in enumerate(body) if i % 2 == 0)
+ 2 * sum(val(c) for i, c in enumerate(body) if i % 2 == 1))
return str((10 - total % 10) % 10) == chkThese implementations pass the reference vectors used throughout this document: s10_valid("RA473124829US"), ups_valid("1Z999AA10123456784"), and usps_valid("9400111899223197428497") all return True, and flipping any single digit makes them return False.
A tracking number tells you less than you would hope, and it is worth being honest about the limits:
A passing check digit proves the number is well-formed, not that a shipment exists. Anyone can construct a syntactically valid S10 or UPS number; only the carrier's system knows whether it was ever assigned to a real parcel.
Shape overlaps are real and unavoidable. Plain numeric formats such as FedEx, DPD, Evri, and Canada Post domestic share length ranges, so a purely offline detector cannot always give a single answer. Treat detection as narrowing, not deciding.
Carriers change and extend their formats. New services, acquisitions, and label-system upgrades introduce new prefixes and lengths over time. The patterns here reflect the common formats in circulation but are not a frozen specification.
When you actually need the delivery status rather than just the format, resolve the number against the carrier, or against an aggregator like InstantParcels that queries the right carrier for you regardless of which format the number turns out to be.