Yeast Starter Calculator
Enter your starting cells, starter volume, and gravity — get the exact new cells grown, grams of DME to add, inoculation rate, and final cell density. Built on the gravity-aware grams_DME = 2.72715 × points × volume_L Braukaiser growth model by a former New Belgium yeast specialist.
grams_DME = 2.72715 × points × volume_L · IR = start_B ÷ grams_DME · growth (B new cells/g DME) depends on IR and agitation — stir plate scales down from 1.4 B/g as IR rises; shaking and still are flat at 0.62 and 0.40 until IR > 3.5, where growth stops.
Size a starter that actually grows the cells you need.
A starter isn't just "yeast in wort" — it's a growth reactor, and growth depends on how much extract you feed it relative to how many cells you're already carrying. Pitch too little DME for the cells you're starting with and you barely grow at all; pitch a huge starter you don't need and you've wasted extract and time. This tool runs the numbers so you know, before you build it, how many new cells a given volume and gravity will actually produce — not just "make a 2 L starter and hope."
Everything runs client-side. Nothing is stored, nothing is sent — change any input and the result updates live.
The starter growth formula, explainedThe Braukaiser (Troester) model.
Extract mass first: grams of DME from volume and gravity
The model starts from a simple empirical fact: about 100 g of DME per liter produces a wort around SG 1.037 — the "10:1 rule" homebrewers already use to size a starter. Braukaiser's constant reproduces that exactly, so any combination of volume and gravity converts cleanly into grams of DME, the real quantity the yeast are metabolizing.
points = (SG_starter - 1) × 1000
# 2.72715 g DME per gravity-point per litre — reproduces the 10:1 rule
grams_DME = 2.72715 × points × volume_L
Inoculation rate: how crowded is the starter, really
The key ratio isn't cells-per-liter, it's cells per gram of extract — the inoculation rate (IR). A small pitch into a lot of DME has a low IR and a lot of headroom to grow; a big pitch into a small starter has a high IR and is already crowded before it starts.
startB = startingCells_billion × (viability / 100)
# billion cells already present, per gram of DME available
IR = startB / grams_DME
Growth per gram of DME depends on IR and agitation
This is where the model earns its keep: growth isn't a flat multiplier, it's a curve. On a stir plate, a starter with IR below 1.4 grows at a strong flat rate of 1.4 billion new cells per gram DME; between IR 1.4 and 3.5 that rate falls off linearly as the starter gets more crowded; above IR 3.5 there's no meaningful growth left — you're maintaining, not growing. Shaking and still (no agitation) starters don't get that low-IR bonus at all — they're flat at 0.62 and 0.40 B/g respectively, and both also stop producing new cells past IR 3.5.
stir plate: IR < 1.4 → growth = 1.4
1.4 ≤ IR ≤ 3.5 → growth = 2.33 − 0.67 × IR
IR > 3.5 → growth = 0
shaking: IR ≤ 3.5 → growth = 0.62 (flat) ; IR > 3.5 → growth = 0
still: IR ≤ 3.5 → growth = 0.40 (flat) ; IR > 3.5 → growth = 0
new_cells = growth × grams_DME
final_cells = startB + new_cells
final_density_B_per_mL = final_cells / (volume_L × 1000)
100 B pack, 2 L starter at 1.037, stir plate. points = 37. grams_DME = 2.72715 × 37 × 2 ≈ 202 g.
IR = 100 ÷ 202 ≈ 0.50 — below 1.4, so growth stays at the flat stir-plate rate of 1.4 B/g.
new_cells = 1.4 × 202 ≈ 283 B. final_cells = 100 + 283 ≈ 383 B. final_density = 383 ÷ 2000 ≈ 0.19 B/mL.
When the standard starter number is wrong.
Starter math has a credibility problem: run the same starter through two "standard" calculators and you can get two different answers. Here's where that happens and what to actually trust.
The models disagree — and neither is gospel
This tool uses the Braukaiser (Troester) model, which scales with extract mass — a 2 L starter at 1.060 grows differently than the same 2 L at 1.037, because there's more DME to work with. The popular Mr. Malty / White curve is gravity-independent: it assumes a fixed 9 °P extract regardless of what you actually pitched, and is a function of inoculation rate alone. For the same starter, the two models can and do predict different final cell counts — sometimes sizing a wildly different volume for the same target. Treat any starter figure, from this tool or any other, as ±10–20%, not an exact promise. Build in margin, don't chase the decimal.
Stir plates aren't magic
Conventional wisdom says a stir plate multiplies your yield. The sourcing behind that bonus is thinner than the confidence people put in it — Brewer's Friend's own documentation notes the stir-plate and shaking bonuses bolted onto White's curve have no citation in White's original study, and Troester's own data raises the question of whether stepping up is even necessary once you're already stirring. A stir plate keeps CO₂ off the yeast and O₂ available, which genuinely helps — but don't assume it's rewriting the growth curve by some specific multiplier you read once.
Bigger isn't linearly better
Look at the growth curve again: as IR rises, growth per gram of DME falls. That means doubling a starter's extract doesn't double your new cells once you're mid-curve — and past IR 3.5, more DME barely moves the final count at all. If your target needs a lot more cells than one starter step can efficiently deliver, don't just build one bigger starter and hope; step up in stages instead, each step targeted at a low, efficient IR. Plan a step-up schedule →
Viability feeds forward into a smaller starting base
The starter doesn't grow from the number on the pack — it grows from startB, the viability-adjusted cell count. A three-month-old liquid pack isn't 100 B, it's closer to 40 B, and that smaller, weaker base changes both your IR and your final yield. Set the viability field honestly instead of assuming "fresh," or the whole calculation is quietly wrong from the first step. Estimate real viability from pack age →
Gravity has a sweet spot: 1.036–1.040
More extract isn't automatically better here either. Pushing starter gravity above about 1.045 stresses the yeast — higher osmotic pressure works against the population you're trying to grow, without a matching improvement in cell yield. Stay in the 1.036–1.040 range (the classic ~1.037 DME starter) and let volume, not gravity, do the scaling.
Notes from the labAt New Belgium, nobody built a starter and just trusted the size. We built it, we counted it, and the count told us whether the model matched reality that week. Two starters that looked identical on paper — same pack, same DME, same stir plate — could finish 20% apart in cell density because one pack was three weeks older than the other, or the gravity was read off a thermometer nobody temperature-corrected. The number a calculator gives you is a plan, not a result. Build the starter the math says to build, then verify it with a count before you trust it in the fermenter.
Track starter yield against what actually pitches clean.
A starter size is a prediction. The value shows up when you log the count you actually get against the model's estimate, batch after batch — and see whether your process runs above or below the standard curve. That's what a QC dashboard does automatically.