When to harvest cannabis: trichomes, pistils and timing

Cut when most trichome heads on mature calyxes have turned cloudy and 10–20% have turned amber, checked separately on the upper, mid and lower thirds of the canopy, never on sugar leaves alone. That is the default most photoperiod growers work to. Everything else on this page is about the parts that single sentence hides: why pistils are a weaker check than trichomes, why the breeder's stated flowering time is a starting estimate and not a countdown, and why the honest answer to "which day" is usually two days, not one.

Default trichome target

Mostly cloudy, 10–20%amber

grower consensus, not a physiological standard

Typical harvest window

5–10days

from first-ready zone to last

Upper-to-lower canopy lag

3–10days

grower estimate; light-driven

Breeder time vs actual finish

±1–2weeks

grower-reported spread around the stated figure

Three signals, and how much each one is worth

Growers lean on three cues to decide when to cut, and they are not equally trustworthy. Ranking them changes how you should spend your attention in the last two weeks of flower.

The rest of this guide works through each of those in the order you should actually use them: trichomes first, pistils as a cross-check, breeder time as the calendar you started counting from.

Trichome head colour: the primary signal

A trichome is a tiny stalked gland on the surface of the calyx (the swollen part of the flower under each pair of pistils) that manufactures and stores the flower's resin. Trichomes begin small and glassy-clear, swell and turn milky white or "cloudy" as the gland fills, and finally turn amber as the resin ages on the plant. Cannabis produces several trichome types, but the stalked-capitate trichomes on the calyx are the ones growers read, because they carry the largest, most visible heads and the clearest colour change ^[1].

That development is not instant, and it is not the same rate on every plant. One indoor trial tracking two cultivars side by side through an 8-week flowering run found the proportion of mature, stalked-capitate trichomes rising through the flowering period to a peak around week 8, with trichome numbers differing significantly between the two cultivars at every stage checked ^[1]. In other words, "week 7" means something different on two different cultivars in the same tent, which is one reason a fixed calendar date is a weak substitute for actually looking.

Commonly used trichome targets, by what a grower is aiming for:

Two practical points follow from how trichomes actually mature. First, check mature calyxes, not sugar leaves: leaf trichomes are smaller, sparser and mature days ahead of the ones on the flower itself, so a leaf that reads "ready" tells you nothing reliable about the bud underneath it. Second, check more than one cola. The next two sections cover why.

Pistil colour and retraction: a useful cross-check, not the main read

The pistils are the pair of hair-like stigmas that emerge white from each calyx to catch pollen; as the flower matures they darken toward orange, red or brown and curl in against the calyx. Watching that happen is the oldest harvest cue there is, because it needs no magnification, and it remains genuinely useful — just not as the primary signal.

Illustrative progression; percentages are the grower-practice range described in the article, not a measured schedule.

As a rough secondary cue, growers commonly use 60–70% or more of a cola's pistils darkened and curled as a sign that the flower is approaching the same maturity window the trichomes describe. The most direct test of how well that actually tracks cannabinoid development comes from a 2025 study that scored stigma colour on a four-stage scale, from no amber pistils through to fully amber, and matched it against measured cannabinoid concentration across 25 cannabis genotypes. Cannabinoid concentration peaked between the "mostly amber" and "fully amber" stigma stages in 22 of the 25 genotypes; the remaining genotypes peaked earlier, at the "no amber" or "some amber" stages ^[2]. The authors' own conclusion is a fair summary for growers: "mostly amber" pistils are a useful rule of thumb for most cultivars, but a genuinely reliable call needs the trichomes, and ideally needs you to know how your specific cultivar behaves.

Pistils have a second weakness trichomes mostly avoid: they darken and curl for reasons that have nothing to do with ripeness. Heat stress, physical knocks, unpollinated flowers left too long, and even normal ageing on lower, shadier colas can all darken pistils early. Use pistil colour to decide whether it is worth getting the loupe out, not to decide whether to cut.

Where you check changes what you see

A single plant does not ripen as one unit. Upper colas sit closest to the light, receive more photosynthetically active radiation and finish first; mid-canopy colas lag behind them; the lowest colas and any "larf" growth shaded by the canopy above can still be running mostly clear trichomes while the top of the plant is well into cloudy-and-amber.

This is not a coincidence of geometry alone. A meta-analysis pooling multiple cannabis cultivation trials found that cannabinoid content per unit canopy area rose with light intensity, and proposed that cannabinoid and resin production is, in part, a response to light stress on the tissue producing it ^[3]. A shaded lower cola is not just slower to fill out; it is developing under a genuinely different light environment to the cola fifteen centimetres above it, which is the physical reason the trichome check, not just the calendar, needs to happen zone by zone.

Illustrative; trichome mixes match the worked example in the article, day offsets are a grower estimate, not a measured trial result.

That spread is large enough to change the decision, not just the numbers. Checking one cola near the top of the plant and applying the result to the whole harvest routinely overstates how ready the lower half actually is. The fix is cheap: sample the upper third, the mid canopy and the lower colas separately, on more than one plant if you are running more than one, and read them as three data points instead of one.

A worked example: two dates, not one

A photoperiod cultivar sits at week 8 of flower, one week inside the breeder's stated "7–9 week" flowering time. A loupe check across the canopy gives:

• Upper colas: 80% cloudy, 15% amber, 5% clear
• Mid canopy: roughly 65% cloudy, 5% amber, the rest clear
• Lower colas and larf: mostly clear, a handful of cloudy heads starting to appear

The upper canopy already sits inside the common default target from the table above. The lower canopy does not, and forcing a single cut date means choosing between an early, underdeveloped lower half or an over-ripe upper half.

1. Cut the upper and mid canopy nowDay 0

   These colas are already at the common target and gaining little by waiting; every extra day is a day of amber risk with no offsetting benefit.

2. Leave the lower canopy on the plant

   With the top growth gone, more light reaches what is left, and lower colas typically finish measurably faster than they would have under the intact canopy above them.

   Warning Removing the top canopy changes the lighting the lower colas receive, which is the point.
3. Re-check in 5–7 daysDay 5–7

   Sample the lower canopy again rather than guessing from the calendar. It should now be approaching the same cloudy-and-amber mix the top canopy showed at the first cut.

4. Take the second cutDay ~7

   Close the harvest once the lower canopy reaches the target, rather than leaving it to catch up with an upper canopy that is by then well past it.

This is a staggered harvest: two or more cuts on the same plant, timed to each zone's own maturity rather than to the plant as a whole. It costs an extra day of harvest labour and an extra day the room stays occupied; it buys a materially more even final product than one cut taken at whatever compromise point looks least wrong. For a fuller treatment of when staggering is worth that cost and when it is not, see the staggered harvest guide.

Extending the wait on the lower canopy carries a real cost of its own in humid rooms: dense, still-wet colas left an extra week are exactly the conditions botrytis (bud rot) favours, so a longer second pass needs tighter humidity control, not just patience.

Why the breeder's number is a starting estimate, not a countdown

Seed and clone vendors publish a flowering time, commonly a range such as "8–9 weeks", measured under their own trial conditions. Two things push a real grow's finish date off that number, sometimes by a week or more in either direction.

The first is genuine biology: environment measurably changes how a cannabis plant's flowering programme runs. Even small differences in day length change flowering behaviour: in a controlled trial across 27 hemp cultivars, flowering in some oil-type cultivars was delayed by 7–8 days when the photoperiod (hours of light, not darkness) exceeded 14 hours rather than 12, and a photoperiod difference as small as 15 minutes, from 13 hours 45 minutes to 14 hours, delayed flowering by 5–13 days in the most sensitive cultivars ^[4]. That trial used industrial hemp rather than drug-type cultivars and looked at time to flower onset rather than time to full maturity, so treat the exact numbers as illustrative rather than a figure to plan around: the underlying point still transfers, that Cannabis sativa's flowering timeline is genuinely sensitive to small environmental differences, not just to which seed packet you bought.

Temperature works the same way in grower experience, though the effect has not been rigorously sized in controlled cannabis trials: warmer late-flower temperatures are widely reported to speed development along, and cooler ones to slow it, which is why growers commonly report their own finish date sliding by roughly one to two weeks around the breeder's stated figure depending on how their room runs relative to the breeder's (grower practice; not independently verified for cannabis).

The second factor is genotype variation the breeder's own number cannot capture: a meta-analysis of cannabis yield studies found significant interactions between cultivar and growing conditions across the dataset, meaning the same environmental change affects different cultivars by different amounts ^[3]. Two plants grown from seed under the same "8-week" label are not guaranteed to finish on the same day even in the same room.

None of this makes the breeder's number useless. It is exactly the right number to start counting from, and the right point to start checking trichomes rather than trusting the calendar. For the physiology behind why flowering time moves the way it does, see breeder flowering times vs reality; for how photoperiod itself drives the flowering clock in the first place, see photoperiod cannabis grown indoors.

What changes at commercial scale

A licensed room adds constraints a home tent does not have. The flip date, the day a room's photoperiod switches to 12/12, is usually the anchor the whole batch is scheduled against: expected harvest date, labour rostering and, in track-and-trace regimes, the flower-day count reported to the regulator all key off it. That makes "when is this room ready" a scheduling question as much as a horticultural one.

The sampling problem also scales up rather than away. A single representative plant per room is a common shortcut and a common source of error, for the same reason a single cola is unreliable in a tent: canopy position, airflow and light uniformity vary across a room, so a handful of plants across different zones gives a genuinely more accurate read than one plant checked carefully.

A worked example at this scale: a 300 m² flower room on one flip date reaches week 8 with loupe checks across three zones showing the two zones nearest the lights at 75–80% cloudy and 10–15% amber, and the zone nearest the intake vents still mostly clear with occasional cloudy heads, a pattern consistent with the light-driven lag described above. Cutting the whole room on the scheduled day either under-cuts the ready two-thirds or over-extends them waiting for the lagging zone; a two-pass cut, ready zones on the scheduled day and the lagging zone 4–6 days later, keeps each pass's flower-day count and wet weight tied to its own zone rather than averaging a mixed-maturity room into one batch record.

Wet weight is captured at the point of cutting, plant by plant or by identified sub-batch, and is the traceability seam where individual-plant records typically collapse into a shared batch record — get the harvest-day paperwork and the calibrated scale right before the room is ready, not on the day itself.

The one mistake worth naming directly

The most common harvest-timing mistake is not cutting too early. It is deliberately waiting well past the common default, pushing trichomes toward 40%, 50%, 60% amber or more, because more amber is assumed to mean something better is happening in the flower. Framed as cannabinoid chemistry rather than as an effect claim, what is actually established is narrower and more genotype-dependent than that assumption suggests. As the pistil-stage study above found, cannabinoid concentration was still rising, not falling, through the mostly-to-fully-amber range for most of the genotypes tested ^[2], while trichome heads that have collapsed or darkened at the stalk (visibly past amber, not merely amber) are a different and later stage associated with degraded, drying resin in most genotypes. "More amber" is not a dial that reliably moves cannabinoid content in one direction for every cultivar, and this library does not treat trichome colour as a potency or effect indicator at all: it is a timing cue, read alongside pistils and the calendar, for deciding when a specific flower on a specific plant has finished developing.

If you are not sure whether your own cultivar runs early or late against these defaults, the answer is not to guess further out on the colour scale. It is to check more often in the final two weeks, log what you see against the eventual cut date, and let that record, not a stock percentage, set your target next time.

What to log at every cut

• Date and week of flower at the time of checkingcounted from the switch to 12/12, not from the breeder's stated figure
• Trichome mix by zone: clear, cloudy, amber for upper, mid and lower canopyeven a rough visual estimate beats no record at all
• Pistil colour and retraction as a rough percentagecross-check only, recorded alongside the trichome read
• Cultivar and sourceseed or clone, and supplier or mother plant, so next cycle's target starts from your own data
• Wet weight per zone, plant or sub-batch at the point of cuttingthe single figure every later stage (drying, curing) gets measured against

Before any of this, confirm what your own law section allows for plant count and possession once a harvest turns growing plants into stored flower; Malta's home-grow rules, for one worked example, are covered on the Malta law page. For the reading technique itself, see using a loupe or microscope to read trichomes; for the separate, well-studied question of pre-harvest darkness, see 48 hours of darkness before harvest: the evidence; and for turning a cut date into a harvest day that actually runs on schedule, see planning harvest day.