Thermochemical processes are promising ways for energy valorization of biomass and waste, but suffer from a lack of predictability. In this work, we focus on using model molecules to model the behavior of wet organic residues during hydrothermal liquefaction (HTL), a process used to produce bio-based liquid fuels from wet biomass. Monomeric and polymeric model molecules were used as modelling tools to study HTL of real resources. Experiments with model mixtures and four food processing residues (blackcurrant pomace, raspberry achenes, brewer’s spent grains, grape marc) were conducted at 300 °C, 60 min holding time and a dry matter concentration of 15 wt%. To elaborate model mixtures, four model monomers (glucose, guaiacol, glutamic acid, linoleic acid) and two model polymers (microcrystalline cellulose, alkali lignin) were selected from characterization of blackcurrant pomace. HTL of model mixtures reproduced HTL of blackcurrant pomace with acceptable representativeness, but results showed that model mixtures should include polymers to represent the fiber content of the resource. Results of HTL of model compounds were used to elaborate polynomial correlations able to predict experimental yields as a function of the initial biomass composition. Calculations were within −8.0 to +4.8 wt% of experimental yields obtained by HTL of real food processing residues, showing a good accuracy of the correlations. These expressions also showed good agreement with HTL results reported in the literature for other resources, and could be useful to assess the potential of various kinds of bioresources for HTL.