How Many ICD-10-CM Codes Are There? Counts by Fiscal Year, FY2016 to FY2027

The number depends on what is being counted. For this guide, the CMS/CDC ICD-10-CM release files are the authoritative source. The order file in each release, icd10cm_order_YYYY.txt, provides the machine-readable list used here to distinguish reportable codes from category and subcategory headers: it lists every entry in tabular sequence with a one-character flag, 1 for a code that is valid for reporting, 0 for a header that needs further characters before it can be used.

For FY2027 that file has 98,403 entries. 74,879 carry the valid flag. The remaining 23,524 are headers such as I10's neighbours I11 or I13, which exist in the tabular but cannot appear on a claim without a further character. Both figures are correct; they answer different questions. A coder asking "how many codes could I report" wants the first number. Someone counting the size of the classification wants the second.

The widely quoted "about 70,000 codes" figure is the valid-code count from the year the United States adopted ICD-10-CM: FY2016 had 69,823 valid codes in 91,737 entries. Eleven annual updates later the valid-code count has grown by 5,056 (7.2 percent) and the entry count by 6,666 (7.3 percent).

Each fiscal year begins October 1 of the prior calendar year. "New" and "removed" are counted the way CMS counts them in its release announcements: a code is new when it is valid this year and was absent or a header last year; it is removed when it was valid last year and is now absent or has become a header.

* The FY2024 order file in the CDC archive includes the April 1, 2024 mid-year additions (the W44 category is present), so its new-code figure is higher than the October 2023 announcement. Every other year is the October release.

Two years stand out. FY2017, the first update after the freeze that preceded ICD-10-CM adoption, added 1,974 valid codes. FY2023 added 1,176, most of them in the injury and external-cause chapters and in the new dementia and social-determinant expansions. FY2022 was the smallest update at 168. FY2027's 190 new valid codes is a below-average year.

The underlying files, the diff for each year, and the FY2027 change lists are published as CSV and JSON in the MedCoder open-data pack (CC0, linked under Sources), so every row above can be checked against the CMS file it came from.

The FY2027 order file has 217 more entries than FY2026, but the valid-code count rises by only 160. The gap is where the two ways of counting diverge, and the FY2027 release shows every mechanism at once.

• 238 entries added, of which 190 are valid codes and 48 are headers. The largest block is T52, where the deleted T52.8X- family is replaced by T52.81- (alkenes), T52.82- (cycloparaffins) and T52.89- (other organic solvents): 51 new entries, 39 of them valid A/D/S codes and 12 of them subcategory headers. New headers such as O00.3, K6A, M67.A and QA1 are entries but not codes.
• 21 entries deleted, of which 15 were valid codes. The whole T52.8X- family (17 entries: the T52.8X header, four subcategory headers, twelve A/D/S codes) and S23.420 with its three encounter codes.
• 15 valid codes converted to headers. These stay in the file, so they are not "deleted" entries, but they stop being reportable. D69.1 becomes the header for D69.11 and D69.19; I42.0, I42.8 and I49.8 gain children; M86.8X1 through M86.8X8 each gain laterality codes; M72.2, Z68.1 and Z87.890 likewise.

So: new valid codes 190; removed valid codes 15 deleted + 15 converted = 30; net +160, which is exactly 74,879 minus 74,719. Counted as entries instead: 238 added, 21 deleted, net +217, which is 98,403 minus 98,186. Neither count is wrong. Only the second one, though, tells a coder that D69.1 stops being reportable from October 1.

1. Download the year's "code descriptions in tabular order" zip from CMS (FY2026 onward) or the CDC/NCHS FTP archive (earlier years) and extract icd10cm_order_YYYY.txt.
2. Count the lines: that is the entry total. Count lines whose character at position 15 is 1: that is the valid-code total.
3. For year-over-year change, key both files by the code in columns 7 to 13 and compare: present in the new file only = added entry; present in the old file only = deleted entry; valid now but not valid before = new valid code; valid before but not valid now = removed valid code.
4. Use the order file, not the tabular XML, to decide which codes are reportable. The XML states a category's seventh-character table once and leaves the combinations implicit, so counting from it means guessing, and guessing has produced codes CMS never defined.

The release history pages on this site are built by exactly this diff and list every code in each bucket.

• Quoting the entry total as the number of reportable codes. About a quarter of the entries are headers.
• Quoting "68,000" or "70,000" as the current count. That was FY2016. Nearly 5,000 valid codes have been added since.
• Treating "deleted" and "no longer valid" as the same set. Fifteen FY2027 codes remain in the file and are still not reportable, because they became headers.
• Comparing an October release file against a mid-year (April 1) file and reporting the difference as the annual update.
• Counting ICD-10-PCS or HCPCS in the same figure. ICD-10-PCS is a separate procedure classification with its own release and its own counts.

The FY2016 to FY2025 files are the archived icd10cm_order_YYYY.txt releases on the CDC/NCHS FTP site. The FY2026 and FY2027 files are the CMS "code descriptions in tabular order" releases. Every count on this page is a mechanical line count or diff of those files, re-run on 2026-09-07 by the same script that produces the open-data pack; no figure was taken from a secondary source or from memory. CMS's own announcement figures, where they are published, agree with the valid-code basis above (for example FY2023: 1,176 new; FY2026: 487 new).