Study compares freeze, oven and humidity-controlled drying of cannabis flower
A peer-reviewed laboratory study in the Journal of Cannabis Research found the drying method affected terpene content in cannabis flower far more than its cannabinoid content.
Research Digest — peer-reviewed laboratory study (Article in Press)
A study accepted for publication in the Journal of Cannabis Research compared three ways of drying harvested cannabis flower — freeze drying, oven drying at 40 °C, and slow drying in a chamber controlled for vapour pressure deficit (VPD) — and found the method affected terpene content far more than cannabinoid content.
The study is a laboratory experiment on plant material, with no human or animal participants. Researchers from Aristotle University of Thessaloniki, Greece's Hellenic Agricultural Organization DIMITRA and the company IA-AGRO dried flower from one high-THC cultivar, all cloned from a single plant, in three replicate batches per method. The paper is peer-reviewed and appears as an Article in Press, so the final edited version may differ. It states the work was partly funded by the company HECAGEN; the authors declared no competing interests.
The study found freeze drying reached the 12% moisture target in about 11 hours, against about 35 hours for oven drying and six to seven days for VPD drying. Freeze-dried samples kept more green colour and intact trichomes, but yielded almost 40% less essential oil and had lower THCA (19.1%, against 21.8% in fresh flower). VPD drying kept the most essential oil (2.5%, similar to fresh flower), and THCA after oven and VPD drying matched fresh material.
The authors suggest VPD drying gave the best overall balance, and that the choice should depend on which qualities a producer prioritises. They say storage and curing effects still need study.
Source: Journal of Cannabis Research — Post-harvest drying determines cannabinoid and volatile composition in pharmaceutical cannabis (Cannabis sativa) (accepted 25 September 2026; Article in Press, online publication date not stated on the document; DOI 10.1186/s42238-026-00514-9)