The extraction efficiency of aflatoxin B1 (AFB1) in cattle feed containing nine adsorbents (ADSs) was investigated using two organic/aqueous solvents composed of methanol/water (80/20 v/v; MeOH) and acetone/water (85/15 v/v; AC). Samples were obtained including a highly AFB1-contaminated (HC) and a low-level AFB(1)-contaminated (LC) feedstuff (15.33 and 7.57 microg kg(-1), respectively), nine ADSs (four clay minerals; one yeast cell wall-based product; one activated carbon and three commercial ADS products) at two different levels of inclusion (10 and 20 g kg(-1)). After solvent extraction and immunoaffinity column clean-up, all samples were analysed for AFB1 by high-performance liquid chromatography (HPLC) with fluorescence detection. For each contamination level (HC and LC), the data obtained were analysed using a factorial arrangement in a completely randomized design. Means were compared with the correspondent controls using the Dunnett's test. No statistical difference was found in AFB1 levels of feedstuffs not containing ADSs when extracted with AC or MeOH, even if numerically higher values were obtained with AC. A dose-dependent effect (p < 0.01) of ADSs inclusion was observed on AFB1 recoveries that were lower when the higher ADS level (20 g kg(-1)) was included in the HC and LC feedstuffs. Higher AFB(1) recoveries were obtained using AC compared with MeOH, both in HC (75.0% versus 12.0%, respectively) and in LC (84.0% versus 22.8%, respectively) ADSs containing feedstuffs. However, when the activated carbon and the sodium bentonite were included in feeds, lower AFB1 concentrations with respect to control values (p < 0.001 and <0.05, respectively) were obtained also using AC. The data obtained in this study indicate that routine use of the MeOH solvent for AFB1 analysis of unknown feedstuffs, can produce misleading results if they contain an ADS.