How tall can a sign actually be? 20,000 codes, one distribution

“How tall can this sign be?” is the single most common question a sign contractor fields before a job even gets bid, and it is usually answered with whatever number the last few jobs happened to use. We have a companion article, our monument sign height guide, that covers how height is measured and why it varies by zone. This article covers something different: the actual statistical distribution of height limits across every ground-sign height rule in our corpus, and what that distribution tells you about how much a single “typical” figure can be trusted.

The scale of the dataset

Our corpus contains 20,188 ground-sign maximum-height rules — an exact count from a direct database query, not an estimate. Of those, 12,281 state a number in feet and 7,759 carry no parseable unit at all, usually because the ordinance expresses the limit in prose or ties it to another dimension. The percentiles below are computed over all 12,281 feet-denominated rules, not a sample.

That is a correction as well as a method note. An earlier version of this article built its distribution from a 3,000-row sample paginated in primary-key order, which is an oldest-rows-first sample rather than a representative one, and it reported a median of 12 ft and a 75th percentile of 20 ft. Measured across the full population the median is 14 ft and the 75th percentile is 25 ft — the sample was systematically low. The same fault, and what it did to a different statistic, is documented in what a sign code doesn’t tell you.

The distribution, in feet

A few observations worth calling out individually:

• The floor is barely above ground level. At the 10th percentile a code allows just 5 ft — a reminder that some jurisdictions regulate very low-profile signage (small identification or directional signs) with the same rigour as a full monument sign.
• The middle of the market sits between 8 ft and 14 ft. The 25th percentile (8 ft) and the median (14 ft) bracket what most people mean when they picture a “standard” monument or low pole sign. If you had to guess one number for a code you have not looked up yet, 14 ft is the statistically defensible guess — but it is a guess for only half the corpus.
• The top quarter of codes allow well above the median. The jump from the 75th percentile (25 ft) to the 90th (35 ft) is steep, and the 99th percentile is 100 ft. Codes in this range are allowing pole-sign-style heights rather than monument-style heights, and treating them as outliers rather than a real, sizeable slice of the corpus would be a mistake.
• The spread from p10 to p90 is sevenfold. Five feet at the low end of the “normal” range to 35 ft at the high end means a single flat assumption about sign height is wrong for the majority of jurisdictions, in one direction or the other. That is the same conclusion, reached from a different angle, as why two towns in the same county can have opposite sign rules, which measures how far apart those limits sit between governments that share a border.

A real limitation, disclosed

The single largest recorded value in the full population is 10,000 ft, with a unit of “ft.” No municipal sign ordinance regulates sign height in the range of commercial aircraft cruising altitude — that is an obvious extraction error, almost certainly a misplaced figure such as a parcel number pulled into the height column. It is not alone: 31 of the 12,281 feet-denominated rules record a limit above 200 ft and 17 record one below a foot. Together that is 0.4% of the population, and it is why the table above reports percentiles rather than a minimum and a maximum — percentiles computed over 12,281 rows cannot be moved by a few dozen bad points, and a stated range can be moved by one.

More broadly, 3,075 of the 20,188 height rules — 15.2% — carry a human-review flag. That is well above the 9.65% corpus-wide rate, which makes sense: height is one of the attributes most often stated with a qualifier (“measured from the crown of the adjacent roadway,” “except for freeway-oriented signs”) that the extraction cannot fully capture as a single number. We are naming that rather than quietly filtering it, because any extraction pipeline working across tens of thousands of ordinance documents produces artifacts, and pretending otherwise would misrepresent the limits of the data.

Why the distribution matters more than a single number

Height limits, more than almost any other sign-code attribute, get flattened into a single rule of thumb in casual industry conversation — “monument signs are usually 6 to 8 feet.” That range sits between the 10th percentile (5 ft) and the 25th (8 ft) of this distribution, so it describes roughly the bottom quarter of the corpus accurately and leaves out the other three-quarters, where codes are more permissive. A bid or a fabrication drawing built around the low end of the range will be needlessly conservative in a majority of jurisdictions, and one built around the high end will be non-compliant in a majority of jurisdictions. Neither error is cheap: under-height signage loses visibility value the client paid for, and over-height signage gets flagged at inspection and has to come down.

For a deeper look at how height is actually measured — grade to sign face, grade to top of structure, average grade on sloped sites — and how setback and height trade off against each other in many codes, see our monument sign height guide. That article covers the mechanics; this one is about the shape of the requirement across the whole corpus.

How to check the height limit for your project

A report shows the maximum height for the applicable zoning district and sign type, cited to the ordinance section, so you are working from the actual local requirement rather than a corpus-wide percentile. If it shows no extracted value, the municipality’s code either is not yet in our database or does not state a height limit explicitly — which, given that 7,759 of these 20,188 rules carry no parseable number, is a real possibility worth a call to the permit counter. Creating an account is free.