<p>This study explores the feasibility of using headspace solid-phase microextraction (HS-SPME) combined with gas chromatography-mass spectrometry (GC-MS) to analyse the volatilome of human melanoma A375 cells. The goal was to identify and characterise the volatile organic compounds (VOCs) and monitor alterations induced by treatment with thapsigargin (TG), a drug known to disrupt calcium homeostasis, thereby inducing endoplasmic reticulum stress and ultimately triggering cell death. Reproducibility of experimental conditions is a major issue in biological experiments, which presents intrinsic variability of the samples to be analysed. In our case, initial analysis revealed a significant batch effect, accounting for 56.88% of the total variance. To address this, external parameter orthogonalisation (EPO) was applied, which successfully reduced the batch variance to just 0.16%. After this correction, the treatment factor became the dominant source of variation, explaining 47.12% of the total variance with strong statistical significance (<i>p</i>-value = 0.001). A supervised classification model using partial least squares discriminant analysis (PLS-DA) was developed and validated to characterise the differences between treated and untreated cells. The model achieved a mean overall accuracy of 92.21% and an area under the curve (AUC) of 0.974, indicating excellent discrimination between the two classes. The robustness of these findings was confirmed by repeated double cross-validation and permutation testing, which showed that the model’s predictive ability was not due to random chance. The results demonstrate that TG treatment induces a reproducible and highly discriminant volatilome signature in A375. This suggests that VOCs could potentially serve as biomarkers for monitoring cellular responses to drug treatments.</p> Graphical abstract <p>Graphical abstract created with BioRender.com</p> <p></p>