A rapid analytical screening method allowing simultaneous analysis of few halogenated persistent organic pollutants (POPs) in water samples at ultra-trace levels was developed. Two-metre long capillary traps with an inner diameter of 320 μm, internally coated with 1.2 μm thick 5%-diphenyl-95%-dimethylpolysiloxane or 14%-cyanopropylphenyl-86%-dimethyl-polysiloxane stationary phase, were used to extract some chlorinated pesticides (aldrin, heptachlor, heptachlorepoxide, dieldrin, endrin, 4,4′-DDE, α-endosulfan) and polychlorobiphenyls (PCBs 1, 15, 44, 77, and 180) from aqueous media. Water samples were pushed through the traps at constant velocity using nitrogen. Every trap was installed in a GC oven and connected in series to an analytical column (25 m long, 320 μm I.D. of CP-Sil 8 CB with a 1.2 μm film thickness) installed in a second GC equipped with an ECD detector (GC-GC tandem system). The capillary trap in the first GC was quickly heated from 50 to 280 °C to focus the retained analytes into a narrow zone between the trap outlet and the analytical column inlet, temporarily kept at room temperature outside the first GC oven. After moving this zone inside the oven of the second GC, focused solutes were thermally desorbed and separated into the analytical column by programming the temperature of the second GC. Extraction recovery was always greater than 70–80% and nearly quantitative for most of the analytes. The sorptive properties of the two stationary phases were independent of sample volume and velocity but were slightly influenced by their polarity. The latter effect was used to investigate the possibility of fractionating the two classes of compounds. For this purpose, a capillary trap containing 14%-cyanopropylphenyl-86%-dimethylpolysiloxane was connected in series to a second capillary trap containing 5%-diphenyl-95%-dimethyl-polysiloxane immediately before the trapping step. Adsorbed solutes were fractioned between the two traps by eluting 0.5 mL of water–methanol 40 : 60 v/v. Most of the tested compounds were retained exclusively by one of the two stationary phases. Overall, the proposed method proved to be practically insensitive to laboratory contamination, reproducible, and suitable for the determination of halogenated POPs at trace levels (LoD in the range 5–50 pg L⁻¹).