The method studied uses a combination of a solid-matrix dispersion partition (SMDP) followed by high-performance size-exclusion chromatography on a minicolumn (HPmSEC) of 7.8 mm I.D. for the separation of pyrethroid (PYR) residues from fatty material. The solid-matrix dispersion extraction is carried out by absorbing a fat solution onto an Extrelut-3 cartridge (filled with a macroporous diatomaceous material) and extracting the PYR residues with acetonitrile. Up to 1 g of fatty material can be extracted with 15 ml acetonitrile. The small amount (mean +/- S.D. = 12.4 +/- 5.9 mg) of fatty material which is eluted into the acetonitrile is further removed by HPmSEC. PYR pesticide residues are collected in a 2-ml fraction between 7 and 9 ml, the column being washed up to 24 ml. The two techniques used in series allow a better removal of fat, a greater input of sample and a lower consumption of solvent compared to the sole SEC on macrocolumns, and a lower limit of determination compared to the sole SEC on minicolumns. Recoveries of 9 PYR out of the 14 investigated residues from soya oil were in the range 66-83% at spiking levels ranging 0.49-2.57 mg/kg, while for 6 PYR residues tested at spiking levels in the range 0.13-0.53 mg/kg the recoveries were in the range 80-111%. Recovery of fluvalinate and permethrin could not be calculated due to interferences from soya oil, while lambda-cyhalothrin, esfenvalerate and tralomethrin gave low recovery. The final extract contains small amount (mean +/- S.D. = 2.4 +/- 0.9 mg) of lipid residue and is not completely free from interferences.