The runoff coefficient C is the dirtiest number in stormwater

TL;DR. "Single-family residential" gets you C = 0.30 in ASCE Manual 77, C = 0.50 in TxDOT's hydraulic design manual, and 0.40 in NRCS TR-55 if you back-calculate from a CN of 75. Don't pick one. Pick the value that matches the recurrence interval, the antecedent moisture, and the regulatory intent — and document the choice.

The setup

Q = CIA is the workhorse of small-area stormwater design. Drainage area in acres, intensity in inches per hour, dimensionless coefficient C, and you get peak flow in cfs (give or take Mulvany's 1.008 unit fudge). Civil engineers learn it sophomore year. The math is so simple that nobody questions it.

The problem is that all three letters have hidden uncertainty. A and I you can usually nail down — surveyed area, NOAA Atlas 14 intensity for a given duration. But C is a rabbit hole.

Look at the variation

Same land cover. Different sources. Sample of "single-family residential, suburban":

Source	C value	Note
ASCE Manual 77 (1992)	0.30 – 0.50	Range; midpoint 0.40
TxDOT Hydraulic Design Manual	0.40 – 0.60	Higher because TX storms are flashier
NCDOT (similar)	0.40	Single value
FHWA HEC-22	0.30 – 0.50	For 2- to 10-year events
NRCS TR-55, CN = 70 backed out	~0.20	For a 1-inch storm
NRCS TR-55, CN = 80 backed out	~0.55	For a 4-inch storm

Across mainstream sources, C ranges from 0.20 to 0.60 for the same hypothetical 1/4-acre suburban lot. That's a factor of 3 in peak flow Q for the same I and A.

Why does the variation exist?

Three reasons. They compound.

1. C is rainfall-dependent, but the table doesn't say so

The Rational Method assumes C is constant — fixed for a given land cover. Real watersheds don't behave that way. As storm depth increases, C increases. A 1-inch storm in a residential neighborhood produces almost no runoff (depression storage absorbs most of it, C ≈ 0.15). A 5-inch storm in the same neighborhood saturates everything and runs off above 0.5.

Some agencies acknowledge this with "frequency factors" — multiply C by 1.1 for the 25-year event, 1.25 for the 100-year. ASCE Manual 77 does this. Most state DOT manuals don't.

2. "Single-family residential" means different things

The term covers 1-acre lots in the foothills and 1/8-acre lots packed onto a hillside. Pre-1970 subdivisions had 25–35% impervious cover; modern infill has 50–65%. Calling both "residential" assumes a single midpoint that fits neither.

Better practice: compute a composite C from the actual impervious-vs-pervious split. C ≈ 0.95 for impervious, 0.10–0.30 for pervious. Weight by area. The answer is more defensible than any table value.

3. Antecedent moisture isn't in the equation

A back-to-back storm — second event 24 hours after the first — has a different runoff response than a dry-condition storm of the same depth. The Rational Method has no way to express this; NRCS handles it via Antecedent Runoff Condition (ARC) factors that adjust the curve number.

If your design event is sized off the second peak in a back-to-back storm sequence (common for some embankment dam spillway analyses), the constant-C Rational Method underpredicts. Switch to NRCS or a continuous simulation.

What to actually use

I run a three-value sensitivity check on every Rational Method calc:

• C_low: table low end. Use this to size structures that benefit from undersizing (parking lot bypass, overflow weir invert).
• C_mid: table midpoint, frequency-adjusted. Use as the design value.
• C_high: table high end + 25%, with frequency adjustment. Use this to size structures that benefit from oversizing (inlet capture, pipe).

For a 4-acre suburban site, that's a Q range from maybe 8 cfs at C = 0.30 to 18 cfs at C = 0.60 for the same 25-year intensity. The honest answer to "what's the design flow" is "between 8 and 18 cfs depending on assumptions, and here's why I picked the value I did." That's also the answer that survives a regulatory review.

Free Rational Method calculator with composite-C support: pe-calc.com/tools/rational-method.html

— Michael Flynn, PE
Next issue: time of concentration. Kirpich versus NRCS lag — when they agree, when they disagree by a factor of two, and which one your reviewer is going to ask for.