Antenatal sickle and thalassaemia screening programmes are now established in most high prevalence areas in England. Although screening reliably detects beta-thalassaemia trait, in many cases, results state that alpha-thalassaemia trait cannot be excluded. The detection of couples at risk of a child with hydrops fetalis is one of the aims of the national programme. We, therefore, performed polymerase chain reaction (PCR) for the common alpha-thalassaemia gene deletions to assess the usefulness of this technique in routine screening practice. Between August 2001 and August 2002, of the 5092 women booked at the antenatal clinic, 425 were found to have a mean corpuscular haemoglobin (MCH) <27 pg in the absence of beta-thalassaemia trait; 189 (44.5%) had an MCH <25 pg. All 425 patients underwent PCR analysis for the common deletions: -SEA (South-East Asian), -MED (Mediterranean), -alpha(20.5), -FIL (Filipino), -alpha 3.7 and -alpha 4.2 genotypes. In total, 130 (31%) women were positive for alpha-thalassaemia deletion; 86 (24.7%) were heterozygous for -alpha 3.7, 19 (4.4%) were homozygous for -alpha 3.7, 12 (2.8%) were heterozygous for -alpha 4.2, 1 (0.2%) was homozygous for -alpha 4.2, 11 (2.6%) were heterozygous for -SEA and one (0.2%) was heterozygous for the -MED genotype. Although the detection rate for alpha(+)-thalassaemia was high, a strategy of selective screening using MCH <25 pg and ethnic group (SEA, Middle East or Eastern MED) would have identified all individuals heterozygous for alpha(0)-thalassaemia. Routine molecular screening for all forms of alpha-thalassaemia trait is unjustified in antenatal screening.