International audienceChemometric treatments of near infrared (NIR) spectra were used firstly to understand data structure by principal component analysis (PCA), to discriminate, by partial least squares-discriminant analysis (PLS-DA) regression, French lavender and lavandin essential oils (EOs) samples (n = 160) and the seven varieties (Abrial, Fine, Grosso, Maillette, Matherone, Sumian and Super) and to quantify the main compounds such as linalyl acetate, linalool, eucalyptol and camphor by PLS regression models. The study was carried out over three crop years (2012–2014) to take seasonal variations into account. French lavender and lavandin EOs and their varieties were well classified (100% for lavender/lavandin EOs and between 96 and 100% for varieties) by PLS-DA regression models. The calibration models obtained by PLS regression for the determination of the main compound contents revealed good correlation (≥0.97) between the predicted and reference values. In the case of major compounds including linalyl acetate and linalool, the relative error of prediction (REP) is close to 2.5%. Partial least squares regression vectors allowed us to identify lavandulyl acetate, eucalyptol, linalool, camphor, trans-ocimene,-caryophyllene and linalyl acetate as metabolomic indicators of Fine, Maillette, Matherone, Abrial, Grosso, Super and Sumian varieties respectively. The use of NIR spectra allowed for an improvement in French lavender and lavandin EOs characterization, quality control and traceability