Amphorae found across the Mediterranean in antiquity are thought to have been used to transport high value liquid plant commodities, such as olive oil and wine. Determining their exact contents using chemical analysis is needed to help characterise these important maritime economies and to understand how they changed through time. While the detection and identification of animal fats (dairy products, adipose fats, marine and freshwater products) is now well established using organic residue analysis (ORA), plant products are rarely detected and are difficult to characterise at high taxonomic resolution. Indeed, the main molecular compounds are widespread in the plant kingdom and are highly susceptible to degradation. Wine appears to be an exception but there is little consensus concerning its detection. Despite some undeniable methodological improvements, no widely accepted analytical protocol has been set up and even the specificity of wine-biomarkers has been questioned. Here we applied new extraction and analytical protocols to one hundred transport amphora sherds (5th - 13th century) from Mediterranean contexts, which are believed to have contained predominantly olive oil or wine. A four-step protocol was applied to analyse a wide range of organic compounds. Samples were screened for lipids and phenolic and small acids with GC/MS after solvent extraction, followed by alkaline treatment on the remaining sherd powder. When enough lipids were preserved, acid extraction was performed to deeply investigate the unsaturated fatty acids and perform carbon stable isotope measurements. For samples containing phenolic and small acids, a soft acid extraction was tested to tentatively extract and analyse polyphenols with LC/MS/MS. First screening steps yielded medium quantities of organics in most of the sherds, generally displaying unspecific distributions of lipids. However, the deep analysis of samples containing unsaturated fats reveals some trends in their distribution, suggesting various commodities (different plant oils, possible fish sauce, or mixing of products). Conifer exudates were also identified. Tartaric acid, usually reported to be a wine biomarker, was detected only in very few samples, but numerous other small and phenolic acids were also identified. The diversity of molecular distributions revealed by this study highlights the potential of ORA for identifying plant products absorbed in pottery and the interpretations will be deepened by comparison with modern samples.