Buoyant densities of four Campylobacter jejuni strains were in the range of 1·084–1·087 g ml−1. Milk (3% fat) and chicken skin homogenates had buoyant densities beneath 1·033 g ml−1. Discontinuous buoyant density centrifugation (BDC) in 40% Standard Isotonic BactXTractorTMmedium respectively succeeded in recovering C. jejuni (5×103–5×104cfu ml−1) from spiked milk (3% fat) and chicken skin. NASBA–ELGA detection of C. jejuni (5×102–5×103cfu ml−1) in 12 different food samples using four different sample preparation methods was performed: RNA extraction by heating (filter and non-filter stomacher bag), RNA extraction by BDC (non-filter stomacher bag), RNA extraction by GuSCN–Triton-X-100 lysis and silica-purification (non-filter stomacher bag). BDC succeeded in overcoming inhibition of the amplification reaction except for one of the soft cheese samples. It was noticed that for chicken skin, chicken meat, pork, chicken sausage, turkey meat, milk (3% fat) and skimmed milk a simple heat treatment at 96°C without any additional precautions to prevent inhibition accomplished NASBA–ELGA detection of the pathogen. The use of a filter stomacher bag improved the quality of the NASBA–ELGA detection signal for beef but did not prevent inhibition of the amplification reaction in the cases of ground beef, prepared minced meat, soft cheese and hard cheese. The silica purification of RNA (which was used as a control treatment) accomplished NASBA–ELGA detection of C. jejuni for all of the latter food homogenates.