How to Track Varroa Mite Levels: A Practical Guide for 2026

If you have kept bees for more than a season, you already know that varroa is the problem. Everything else — temperament, supersedure, propolis on your gloves — is downstream of how well you track and respond to mite pressure. The numbers matter, the timing matters, and what you do with the numbers matters more than either one alone.

This is how I track varroa, what I learned by tracking it badly first, and the small set of habits that make the data actually useful.

Why does varroa tracking get neglected?

Mostly because it is the part of beekeeping that is least fun. You are sticky, the bees are loud, and counting mites in a jar is the opposite of finding the queen. The honest answer is that the friction is real and the discipline is hard — and that is exactly why a logging tool that takes the friction out of the recording part pays for itself within a season.

The cost of skipping varroa tracking is high and back-loaded. By the time you notice the colony is in trouble, your treatment window has narrowed, your options are worse, and your winter losses have already been priced in.

What is the right way to count mites?

There are three accepted methods, and they trade off accuracy against speed.

• Alcohol wash. Sample 300 bees from a brood frame, shake in alcohol, count fallen mites. Kills the sample but is the most accurate snapshot. A count of 3 mites / 300 bees is a 1% infestation — the threshold many state extensions flag for action.
• Sugar shake. Same sample size, no alcohol. Bees survive. Less accurate — expect to recover roughly 70–80% of the mites a wash would find — and more prone to operator error if the shake is not vigorous.
• Sticky board (24-hour drop). Place under a screened bottom board, count natural mite drop after 24 hours. Easy and non-destructive but heavily affected by hive size, brood activity, and ambient temperature. Best for trend tracking, not for absolute thresholds.

I use alcohol wash for the cycle-of-record counts (early summer, mid-summer, late summer, before winter feeding) and sticky boards for in-between checks when something feels off.

How often should I check?

A defensible minimum is four checks per year, timed to the colony's life cycle:

1. Spring buildup, when populations are climbing and mites are climbing faster.
2. Mid-summer, before the main flow ends and the brood break that follows.
3. Late summer / early fall, before you start feeding for winter — this is the most consequential check of the year.
4. Pre-winter, after your fall treatment, to confirm it actually worked.

If you have a mite-vulnerable yard or a year of high local pressure, double this. The data point that matters most is the trend between checks, not any single count.

What threshold actually triggers a treatment?

Most extensions converge on a 3% infestation as the action threshold during the brood season — that is 9 mites per 300 bees on an alcohol wash. Below 1%, you can usually defer. Between 1% and 3%, plan a treatment for the next workable window. At 3% or above, treat now.

The reason these thresholds are blunt is that the colony is a moving target. A 2% count in late summer with brood still being capped is a different situation than a 2% count in October with no capped brood — the second is borderline, the first is on the way to disaster.

How do you log this so it is actually useful later?

Three habits that take a logging tool from "I have a notebook" to "I have a record I can act on":

1. Always log the method. A 1% sticky board number and a 1% alcohol-wash number mean different things. The method is part of the data point.
2. Always log the brood state. "Heavy capped brood" or "broodless" is the context that makes a count interpretable a month later.
3. Always log what you did about it. If you caught a 3% wash and decided to defer because the bees were under late-season stress, write that reasoning down. Future-you will not remember.

This is the part where a voice-first inspection tool earns its keep. If you narrate the count, the method, the brood state, and the decision in the same breath, you do not have to remember to come back later and fill in the context. It is in the record because you said it once. (And if you would rather type the inspection on the web portal at the kitchen table, the same fields are there — voice is the fast path at the hive, manual is the comfortable path at the desk.)

What does WhisperBee do with this data?

Three things, in order of importance:

1. Plot the trend on the hive timeline. A single mite count is barely information; the curve from May to October is what tells you whether your treatment plan worked.
2. Promote a treatment action item when a count crosses the standard treatment threshold (around 3 mites per 100 bees). The action item links back to the inspection that triggered it so you can see the count in context.
3. Defer unrelated brood-quality alerts for two weeks after a mite treatment starts. Some products temporarily scatter brood; a "spotty brood" concern in that window is downgraded with the message that you have recently treated, instead of pushing you toward an unnecessary requeen decision.

If you want to see the voice-logging workflow at the hive itself, the features page walks through it end-to-end. If you want to know what the AI actually extracts from a varroa-mention and how to correct anything it got wrong, the FAQ covers it.

What about treatment-free?

If you are running a treatment-free operation, varroa tracking matters more, not less. The hypothesis you are testing is "my bees can manage their own mites" — and you cannot test that hypothesis without a count. Tracking lets you cut your losses on a colony that is failing the test before it crashes and pollutes your other yards.

Where to start if you have not been tracking

Pick one yard. Do an alcohol wash on every colony in that yard this week. Log the count, the brood state, and your decision for each. Repeat in three weeks. That is enough to start a trend line and enough to make decisions on. You can always layer in more rigor later — the cost of starting is much smaller than the cost of not starting.