Resizing

Adjusting a knitting pattern larger or smaller: the percentage method and its limits

Last verified: 2026-11-16 · 7 min read

The pattern is perfect; the size is not. Maybe the largest size stops two sizes short, maybe the child size needs to become an adult size, maybe the sweater must fit shoulders that the grade never imagined. Resizing is a real skill with a real boundary: rectangles scale beautifully, shaped garments mostly don't. This guide walks the percentage method honestly — what it does well, and exactly where it stops being trustworthy.

Key takeaways

Percentage scaling: desired ÷ original × 100, applied to dimensions, then recompute stitch and row counts at your gauge.
It is reliable for rectangles and tubes: scarves, blankets, hats, socks, drop-shoulder bodies.
It is unreliable for fitted garments: armhole depth, neck width and shoulder slope do not scale with chest circumference.
Ribbing and cable repeats must be re-rounded after scaling — an odd stitch count is a broken brim.
Resized garments need more yarn — recompute before buying (the yarn calculator does piece sums).

The percentage method, step by step

Say a hat pattern is written for a 48 cm head and you need 54 cm — an adult head per the CYC head circumference chart. The scale factor is 54 ÷ 48 = 1.125, or 112.5%. Now convert every relevant dimension the same way, at your gauge (say 20 sts and 28 rows per 10 cm):

Cast-on count: original cast-on (96 sts) × 1.125 = 108 — or better, recompute from scratch: 54 ÷ 10 × 20 = 108 sts. Recomputing from dimensions is safer than multiplying pattern counts because it corrects for any gauge difference between you and the designer (see knitting math explained).
Length: to add 2 cm of depth, that is 2 ÷ 10 × 28 ≈ 6 extra rows, inserted before the crown starts.
Check repeats: 108 divides by 4, so a 2×2 brim survives. A target of 51 cm would give 102 stitches — not divisible by 4; choose 100 or 104 and accept 50 or 52 cm.
Crown decreases: for a tube, keep the designer's decrease logic but start from your count; if the pattern says "k2tog around" (halving), your halves are 54 — the geometry of a crown scales, which is why hats are friendly.
Yarn: roughly 12.5% more fabric than the largest size — recompute, don't guess.

The resize calculator automates steps 1–3 for any list of parts, flags repeat-breaking stitch counts, and leaves the judgement to you. For blankets, add blocks rather than scaling stitches: a blanket that is 106% as wide needs one more pattern repeat, not 106% of a repeat.

Where the method breaks: the three big limits

1. Shaped areas do not scale with circumference

Look at any published size chart — the Craft Yarn Council tables on our measurements cheat sheet are a public example: bust grows several inches per size step while armhole depth grows a fraction of an inch and cross-back barely moves. Human bodies scale non-uniformly, and patterns are graded to match. Multiply a whole garment by 112% and you get armholes 12% deeper than any size chart intends — sagging, bagging, and sleeves that don't fit the holes. Percentage scaling a drop-shoulder garment is usually survivable (the armhole is a straight line); a set-in sleeve or saddle shoulder is exactly where the method earns its bad reputation.

2. Ease is not a percentage

If a sweater is designed with 10 cm of ease at a 100 cm chest and you scale everything by 110%, you also scaled the ease to 11 cm — probably fine. But if you scale by 130%, the "same sweater" now has 13 cm of ease plus proportionally deeper armholes, and the silhouette is a different design. Ease wants to be edited in absolute centimetres: decide the finished chest measurement first (body measurement + intended ease), derive that percentage for width, and leave lengths closer to the original.

3. Length is not width

Adults get wider and a little taller; children get dramatically taller and wider in different places. Scaling a child's sweater by 130% in both dimensions produces a long, narrow adult sweater. Scale width and length with separate factors — and when in doubt, compare against published standard measurements rather than pure proportion.

The professional alternative: grade between sizes

What designers actually do when a size is missing is grading: take the two nearest published sizes and interpolate — or blend, body width from one size, armhole and neckline shaping from another. It is more work than multiplication but it respects the garment's architecture. The same logic, with per-measurement differences from a standard size table, is the foundation of sewing grading, covered in sewing grade rules — and knitters can borrow it: use the CYC measurement tables to see how much each landmark "should" change between your size and the pattern's largest, and walk the pattern toward it in those amounts, not one blanket percentage.

A resizing sanity checklist

Recomputed stitch counts are whole numbers and respect ribbing/cable repeats
Armhole depth changed by a centimetre amount, not a percentage
Sleeve stitch counts still match armhole rows for the seam or pick-up ratio
Yarn recomputed for the new dimensions (see the yarn calculator)
A half-size test swatch of the modified numbers lies flat and measures as predicted

When to just not resize

Some patterns fight resizing structurally: heavily cabled panels (repeat math everywhere), intarsia or fair-isle charts (pixel-perfect stitch counts), and anything with "fit" in its name. In those cases the cheapest path is usually finding a similar pattern in the right size — the resizing effort exceeds the knitting effort. Between those extremes, the percentage method with the limits above is a legitimate, well-loved tool. Use it on the rectangle parts, grade the shaped parts by hand, and your modified garment will fit the body you actually have.