Varroa mite counts and treatment thresholds

Varroa is the thing most likely to kill your colonies this year, and the only way to manage it is to count. Not "I saw a mite on a drone", not "the bees look fine": a number, from a repeatable method, written down with a date. Here is how to get that number and what to do with it.

The alcohol wash (the standard)

The alcohol wash is the most accurate field method and the one most thresholds are calibrated against. It costs about 300 bees, which sounds brutal and is a rounding error against what an untreated mite load costs.

Pull a frame of open brood from the brood nest. Mites concentrate on nurse bees, so a sample from the honey supers will undercount. Check the queen is not on the frame.
Shake or brush bees into a tub, then scoop a level half cup into the wash cup. A half cup is about 300 bees.
Cover with 70% isopropyl alcohol (winter windshield fluid also works), close, and swirl steadily for about 60 seconds.
Count the mites that drop through the strainer.
Divide by 3. That is your mites per 100 bees, usually spoken as a percentage: 9 mites in the wash is 3%.

Sugar roll, same sample but with powdered sugar, spares the bees and undercounts somewhat, especially in cool or humid weather. It is a fine method if you use it consistently and know it reads low. Sticky boards (natural drop over 24 to 72 hours) are useful for trend-watching but hard to turn into a treatment decision; do not use board counts against wash thresholds.

Thresholds: when to treat

Commonly used guidance in the US (the Honey Bee Health Coalition's varroa management guide is the reference worth reading in full):

Around 2% (6 mites per 300-bee wash) in spring and early season: treat, or plan to very soon.
3% or more (9+ mites) from mid-summer on: treat now.
Late summer is the count that matters most. The bees raised in August and September are the bees that must live until spring. High mites in August means damaged winter bees, and colonies that "starved" in February often actually died of viruses vectored in August.

Local guidance varies by climate; your state apiarist or extension service may publish adjusted numbers. When in doubt, count more often rather than debating the threshold.

Counting is half; the record is the other half

A single count tells you today. A series tells you whether treatment worked:

Count before you treat. Otherwise you cannot know if the treatment did anything.
Count again 1 to 3 weeks after the treatment ends. A treatment that knocked 9 mites down to 7 did not solve your problem.
Record the product, the dates, and the temperature. Several treatments have hard temperature limits, and several cannot be used with honey supers on. The label is the law; your record is how you prove to yourself you followed it, and how you respect withholding periods before the next harvest.
Rotate active ingredients across the season and across years. Resistance is real and it is managed at the record-keeping level, because nobody remembers what they used 14 months ago.

A workable season rhythm

Count every colony at least four times: early spring, pre-flow, right after harvest, and mid-autumn. Add a count before and after every treatment. That is six to eight numbers per hive per year, each one taking about ten minutes.

Where HiveMind AI helps

HiveMind AI does not replace the wash cup; nothing does. What it does is keep the series honest: mite observations and counts go into inspection records (you can dictate them from the yard, gloves on, and the voice fill turns them into a structured entry), treatments and mite checks are first-class task types with their own reminder lead times, and an inspection that ends with "wash again in two weeks" becomes a linked task once you confirm it, so the follow-up count actually happens. Your whole count history per hive sits on one timeline. Every plan includes it; see plans.