Milk-based media such as the Fed State Simulated Gastric Fluid (FeSSGF) are commonly used in order to simulate the in vivo properties of the fed state stomach. Due to the lack of a specific guideline for standardised sample clean-up in these media, the aim of the current study was to develop an optimum protocol for the extraction and quantification of drugs from the fed state gastric medium based on the APIs' physicochemical properties (lipophilicity, ionisation, aqueous solubility and protein binding). Two different extraction techniques, protein precipitation (PP) and solid phase extraction (SPE) were assessed. A pilot study in six model drugs was performed, with tests using seven different protein precipitation reagents at four different medium:reagent ratios and two drug concentrations as well as different solid phase extraction cartridges and elution protocols. % recovery was analysed using partial least squares (PLS) regression so as to determine the physicochemical parameters affecting the drug percentage recovered. For protein precipitation protocols, drug concentration, selection of protein precipitation reagent and ratio added to the medium significantly affected drug % recovery from FeSSGF (p < 0.05). The same applied for the selection of elution solvent and cartridge type for solid phase extraction. Optimum protocols using MeOH, ΑCN and 10% w/v TCA at a 1:2 FeSSGF:reagent ratio were effective to a larger group of drugs of a wide range of lipophilicity and ionisation, with ΑCN being the most effective in the whole range of log P values (−0.56 to 8.81). Solid phase extraction was proven to be effective for compounds of poor to moderate lipophilicity (log P < 4), with extremely hydrophobic compounds demonstrating lower % recovery values (down to 10% recovery). PLS demonstrated that only for 10% w/v TCA (protein precipitation) and HLB (solid phase extraction) can the effect of key drug physicochemical properties on the final amount of drug recovered be accurately predicted.