A systematic study was undertaken to optimise and compare the performance of different extraction media employing ultrasonic-assisted extraction (USE), for the recovery of nonylphenol (NP) and nonylphenol mono- and diethoxylates (NP1EO and NP2EO, respectively) from different spiked sediments (sand, clay and soil). Dichloromethane (DCM) and ethyl acetate (EtAc) were used alone or with methanol (MeOH). Normal phase high performance liquid chromatography with fluorescence detection (HPLC-FL) was used for separation and quantification. In a first stage, a screening Plackett-Burman experimental design was used as a multivariate strategy to evaluate the effects of three variables (solvent polarity, analytes concentrations and sonication time), at two levels each, on spiked clay. Solvent polarity was found to be the most influential factor, especially on the recovery of NP2EO. In a second stage, based on the screening results, USE time was set at 5 min to evaluate the performances of a 1 : 1 MeOH : DCM and a 1 : 1 MeOH : EtAc mixture on spiked sand, clay and soil. The 1 : 1 MeOH : EtAc mixture led to highly satisfactory recoveries for every analyte, statistically comparable to those yielded by a 1 : 1 MeOH : DCM mixture (NP > 85%, NP1EO and NP2EO > 90%). Due to similar interaction observed between each single sediment and the 1 : 1 MeOH : EtAc extractant, a composite substrate made of sand + soil + clay was spiked in a third stage. Extractant composition was then evaluated in order to find out the minimum volume of MeOH that could be used without loss of efficiency. It was found that 100% EtAc matched our aims (%R > 80 and the lowest use of DCM and MeOH). USE protocol was tested on real sediments. The compounds were quantified by HPLC-FL and the identities were confirmed by gas chromatography-mass spectrometry.