Vials to wells: how many cells your experiment actually needs

The short answer: multiply the published growth area of your well by a seeding density in cells per cm², multiply by your well count, then divide by the usable cells in a vial — which is the labelled count times the counting convention times your own measured post-thaw viability, plus 10 to 20 percent for dead volume. Buy the next whole vial up, and buy the whole order from one lot.

For a 48-well MEA experiment at a typical neuron density, that comes to seven vials of one million viable cells, not five, and the three-vial gap between the naive answer and the real one is where experiments run short at the bench.

The rest of this page is the reference data to do that arithmetic for any format, with every number sourced.

Growth area per well: the published figures

Growth area is the only number in this calculation that is a property of the plastic rather than of your biology, and it is the number people most often guess. Do not guess it — and do not assume two manufacturers agree, because on two formats they do not.

Two disagreements matter.

The 48-well. Corning publishes 0.95 cm², Thermo publishes 1.1 cm² — a 16 percent gap. On a full 48-well plate at 2 × 10⁵ cells/cm² that is 1.44 million cells against 1.58 million, which is the difference between one vial and two at the margin.

The 384-well. Corning publishes 0.056 cm² for a standard 384-well, Thermo publishes 0.084 cm². That is a 50 percent difference, and it propagates directly into how many plates a vial fills. Use the figure for the plate on your bench.

Corning’s own caveat is worth repeating: these areas are calculated from engineering drawings, do not account for moulding variation, and do not account for cells growing up the sides of the vessel, which for some lines adds meaningfully to the real available surface.

Permeable supports (Transwell-type inserts)

If you are working at an air–liquid interface or doing barrier work, the insert area is not the well area.

Note the 24-well insert at 0.33 cm² against the 24-well plate at 1.9 cm². Seeding an insert at a density calculated for the well underneath it over-seeds by a factor of nearly six.

MEA plates: why there is no growth area column

Multi-electrode array plate manufacturers publish well volume, electrode count, electrode spacing and recording area. They do not publish a growth area per well, and you should not compute one from the plate drawing. MEA wells are not flat-bottomed cylinders — they taper to a raised, small-diameter bottom above the electrode array — so an area derived from a well diameter is not a culture surface area.

More importantly, it would not help you, because MEA cultures are almost never seeded by area. The standard protocol across vendors is droplet spotting: a small, high-concentration droplet placed directly over the electrode array, incubated for an hour to attach, then flooded with medium. The number that matters is cells per well, and the vendors publish that directly.

Published cells-per-well figures for MEA

These are from vendor protocols and application notes, for the cell types and plates named. They are the most directly usable numbers on this page.

Three things follow from that table.

Neurons need a density floor to network at all. Below roughly 40,000 cells per well in a 24 or 48-well MEA, you get a culture and no network. This is not a gradual falloff — the published density sweep shows little to no activity at or below 20,000 cells per well and robust spiking at 40,000 and above. Under-seeding an MEA does not give you a weaker result; it gives you no result.

Astrocyte co-culture lowers the neuron count you need and accelerates maturation. Halving the neuron density and adding astrocytes at a 10:1 ratio is a published, cheaper route to a synchronised network than seeding neurons alone at full density. Astrocytes are usually a different SKU from a possibly different vendor, so factor the second order and the second lead time in.

The droplet volume is the constraint, not the well volume. An 8 µL droplet at 6.25 × 10⁶ cells/mL is 50,000 cells. If you need more cells per well you increase the concentration, not the droplet, because a droplet that spreads past the electrode array defeats the purpose.

Seeding densities: the sourced bands

These are the densities published by named suppliers for their own products. Use them as a starting range; every one of these vendors also tells you to optimise for your application.

For the tables below we use a neuron band of 1.0 × 10⁵ to 2.5 × 10⁵ cells/cm² — the span between the lowest and highest published figures above — and a cardiomyocyte band of 1.0 × 10⁵ to 2.0 × 10⁵ cells/cm², which is the published Axol range and contains the FUJIFILM CDI figure.

The neuron band is 2.5-fold wide. That is not sloppiness on anyone’s part; it reflects real differences in protocol route, subtype, co-culture and what the culture is for. It does mean that “how many cells do I need” has a 2.5-fold answer until you fix the density, so fix the density first.

Cells per well, and wells per one-million-cell vial

Calculated on the Corning growth areas above. Substitute your own plate’s figure if it differs.

Neurons, 1.0 × 10⁵ – 2.5 × 10⁵ cells/cm²

Cardiomyocytes, 1.0 × 10⁵ – 2.0 × 10⁵ cells/cm²

Read the “wells per vial” columns as an upper bound, not a plan. They assume every labelled cell in the vial reaches a well. It will not. The corrections below are what turns these numbers into a purchase order.

Two sanity checks against vendor arithmetic. FUJIFILM CDI’s own table gives 2.4 × 10⁶ cells for a 6-well at 2.5 × 10⁵ cells/cm² on their stated 9.6 cm², and 8.0 × 10⁴ for a 96-well at 0.32 cm² — both consistent with the high-density column here. BrainXell states 3.3–4.4 million live cells to fill a 96-well plate, which is 34,000–46,000 cells per well and sits just above our low-density column, because their stated density band starts slightly below 1.0 × 10⁵ cells/cm².

The correction factors nobody puts on the datasheet

Four things sit between the number on the vial label and the number that ends up attached in your wells.

1. The counting convention

A vial labelled “1 × 10⁶ cells” can mean total cells at freeze or viable cells post-thaw. If it is total cells at freeze with an ≥ 80 percent viability specification, your usable count is 800,000, not a million.

This is the largest single correction and the one most often missed. Ask which convention the label uses. See post-thaw viability for why the same vial returns different numbers on different assays.

Correction: × 0.75 to × 0.85 if the label is a freeze-basis count. × 1.0 if it is a guaranteed viable post-thaw count.

2. Your own measured post-thaw viability

Even a viable-post-thaw label is a specification floor measured by the vendor’s method at the vendor’s timepoint. Your own count, at your own bench, after your own thaw, is the one that governs. Count before you plate — every protocol above tells you to, and the reason is that a lot can meet specification and still leave you short if you planned on the label.

The vendor’s method matters here too: trypan blue reads generously on debris-laden post-thaw samples, AO/PI does not. A vendor specification measured by trypan blue and your own count by AO/PI will not agree, and yours will be lower.

Correction: use your measured count, not the label. Plan on the specification floor, not the typical value.

3. Attachment efficiency

Cells that are alive in suspension are not the same as cells attached in a well. Attachment depends on the coating, the medium, the substrate, the density, the handling and the timing. Published protocols are emphatic about this: the entire thaw-to-plate process should not exceed an hour, plates should not be moved or agitated during seeding, and cells must be allowed to settle for ten minutes before the plate is transferred.

The important point for purchasing is that the vendor’s recommended density is a seeding density and already assumes typical attachment. So for the buying calculation you do not correct twice. You correct for attachment only if your target is a specific number of attached cells — for example when a downstream lysis or imaging endpoint needs a known cell number.

Correction: none for the purchase if you are seeding at the vendor’s recommended density. Run a small attachment control at 24 hours and record it, because that number is what predicts your experiment.

4. Dead volume and pipetting overage

You cannot pipette the last of a suspension. Multichannel dispensing, droplet spotting, tube and tip retention, and the need to fill a reservoir all cost cells.

Vendors build this in. One MEA protocol notes that a 48-well plate needs at least 528 µL of cell mix at 11 µL per well, and then recommends preparing 550 µL — a small but deliberate excess. For multichannel work across a full plate, more is required.

Correction: add 10 percent for single-channel droplet work, 15 to 20 percent for multichannel dispensing across a full plate or for anything involving a reservoir.

5. Buy from one lot

Not a multiplier, but the most consequential decision in the list.

Two lots of the same catalogue number can differ enough to break a comparison. If your experiment spans conditions, timepoints or plates, a lot boundary running through the middle of it introduces a variable you did not design and cannot subtract afterwards. Ask the supplier to reserve the whole quantity from one lot, and ask them to say so on the paperwork.

If a single lot cannot cover the order, the fallback is to design the lot boundary into the experiment deliberately — split it across replicates rather than across conditions — and to record which wells came from which lot. That is recoverable. Discovering it afterwards is not.

Worked example: a 48-well MEA experiment in triplicate

The question. I want to run a 48-well MEA experiment in triplicate — 16 conditions, three wells each, human iPSC glutamatergic neurons with astrocyte support. How many vials do I buy?

Step 1 — wells. 16 conditions × 3 replicates = 48 wells. The whole plate.

Step 2 — cells per well. Using the published 48-well MEA co-culture protocol: 120,000 neurons and 20,000 astrocytes per well, spotted as an 8 µL and a 3 µL droplet.

Step 3 — nominal requirement.

• Neurons: 48 × 120,000 = 5,760,000
• Astrocytes: 48 × 20,000 = 960,000

Step 4 — dead volume and pipetting overage. Droplet spotting across a full 48-well plate, 15 percent:

• Neurons: 5,760,000 × 1.15 = 6,624,000
• Astrocytes: 960,000 × 1.15 = 1,104,000

Step 5 — counting convention and viability. Now the answer forks on how the vial is labelled.

Astrocytes at 1,104,000 needed: 2 vials of 1 × 10⁶ viable, or 1 vial if your supplier’s astrocyte vial is ≥ 1.5 × 10⁶ viable — which several are, so check before buying two.

Step 6 — the pack size decision. Look at the third row. Three vials of 2.5 million deliver 7.5 million cells against seven vials of one million delivering 7 million. In our recorded price data, one supplier’s own price per million fell by roughly 46 percent moving from a 1M to a 5M vial. If the larger vial exists for your subtype and genotype, it is very likely both cheaper in total and better matched to the requirement — and it removes six thaw events, which is six opportunities for the process to go wrong.

The counter-argument is real: a larger vial thawed for a 48-well plate is a single point of failure, and you cannot refreeze the remainder. If you have a second experiment queued within the working window, take the large vial. If not, the smaller vials buy you optionality.

Step 7 — lot. Reserve all seven (or three) neuron vials and both astrocyte vials from one lot each, and get it in writing on the order.

The answer: seven vials of one million viable neurons and two vials of astrocytes — or three larger neuron vials if the SKU exists. The naive calculation, 5.76 million ÷ 1 million = six vials, is one vial short before you have thawed anything, and two to three short if the label is a freeze-basis count.

A second worked example: cardiomyocytes in 96-well

The question. A cardiotoxicity screen, 96-well plate, ten compounds at eight concentrations in duplicate, plus controls. One full plate. How many vials?

• Wells: 96
• Density: FUJIFILM CDI’s published figure of 1.56 × 10⁵ viable cells/cm² on a 0.32 cm² well = 50,000 cells per well (which is exactly the figure in their own quick guide)
• Nominal: 96 × 50,000 = 4,800,000
• Multichannel overage, 20 percent: 5,760,000
• Vials at “1 × 10⁶ viable”: 5.76 → 6 vials
• Vials at a 4 × 10⁶ kit: 1.44 → 2 kits, with meaningful spare

Note how differently this reads from a naive 4.8 million ÷ 1 million = five vials. The overage alone adds a vial, and a screening plate that runs short mid-dispense costs the whole plate, not one well.

The calculation, as a formula

For anyone who wants to put this in a spreadsheet:

cells_needed  = wells × replicates × density(cells/cm²) × area(cm²/well)
                                       [or × cells_per_well for MEA droplet seeding]

cells_to_buy  = cells_needed × (1 + overage)
                overage = 0.10 single-channel, 0.15–0.20 multichannel / full plate

usable_per_vial = labelled_count
                  × (1.0 if labelled viable post-thaw, else stated viability)

vials         = CEILING( cells_to_buy ÷ usable_per_vial )

Then check three things the formula cannot: whether a larger pack size is cheaper per million, whether the whole quantity is available from one lot, and whether the media, supplement and coating are on the same order with the same lead time.

Common errors

Using the well area for an insert. A 24-well insert is 0.33 cm², the well beneath it is 1.9 cm². Seeding an insert on the well’s number over-seeds nearly sixfold.

Using one manufacturer’s growth area for another’s plate. Immaterial for 24 and 96-well. Material for 48-well (16 percent) and severe for 384-well (50 percent).

Computing an MEA well area from the plate drawing. MEA wells are not flat-bottomed cylinders, manufacturers do not publish a growth area, and the seeding is droplet-based regardless. Use the published cells-per-well figures.

Under-seeding an MEA to save cells. Below the density floor you do not get a weaker network; you get no network, and the whole plate is wasted. The published sweep shows the cliff between 20,000 and 40,000 cells per well.

Planning on the typical viability instead of the specification floor. The typical value is an observation; the floor is the commitment. Plan on the floor and be pleasantly surprised.

Ordering cells without the media and coating on the same order. They are frequently separate SKUs with separate lead times, and the shortest path to a working experiment is set by the slowest line item. See lead times.

Forgetting the plate itself. MEA plates are single-use and designed for one-time cell plating. A repeat run is a new plate, and MEA plates have their own lead time.

What we would confirm before ordering

• The counting convention on the label, in writing.
• The viability specification floor, the method, and the timepoint.
• The lot-specific seeding guidance on the certificate of analysis — several suppliers state that recommended density varies by lot and put the lot’s figure on the CoA.
• Whether a larger vial size exists for your exact subtype and genotype, and its price per million.
• Whether the full quantity is available from one lot, reserved.
• Whether media, supplement, coating and the plate itself ship on the same order and the same date.

Where we fit

We do this arithmetic as part of normalising quotes, because the counting convention is exactly what makes two prices non-comparable — a supplier quoting a freeze-basis count against one quoting viable post-thaw looks cheaper while delivering fewer usable cells. If you send us the plate format, density, well count and replicate structure, we will come back with the vial count, the pack-size comparison across suppliers, and confirmation on lot availability before you raise a purchase order.