A new analytical method is proposed to determine more than 40 multiclass pesticides in different kinds of processed (whole, skimmed and powdered) and unprocessed (goat and human) milk samples using solid-phase microextraction (SPME). A comparative study between headspace (HS) and direct immersion (DI) was carried out. The effect of milk dilution and the use of acid to reduce the influence of the matrix in DI-SPME mode were also evaluated. DI of the SPME fiber into previously diluted and acidified milk samples achieved the best sensitivity results. Pesticides were determined using low-pressure gas chromatography-tandem mass spectrometry (LP-GC-MS/MS). Both of the selected techniques have been shown to be effective at reduce fat interference and can determine analytes present at very low concentrations (limits of quantification between 0.02 and 1.00 microg L(-1)). Performance characteristics such as linearity, recovery, precision, and lower limits, together with an estimation of the measurement uncertainty using validation data, are presented for each pesticide. All of the pesticides presented recovery rates of between 81 and 110% and precision values lower than 12% (expressed as the relative standard deviation). The overall uncertainty of the method was estimated at three different concentrations (10, 25 and 50 microg L(-1)) and was lower than 25.5% in all cases. The proposed analytical methodology was applied to the analysis of target pesticides in 35 samples: 15 commercial, 3 human and 17 goat milk samples. The metabolite p,p'-DDE was the compound most frequently found in both the breast and goat milk samples, at concentration levels < 20 microg L(-1). However, pesticide residues were not found in any of the other 15 commercial milk samples (skimmed, powdered and whole milk) analyzed.