For over a decade it has been possible to externally sex monomorphic birds using PCR. A major drawback of the protocol developed by Richard Griffiths is that DNA from at least 20 000 cells is needed (Griffiths et al. 1996 Proc. Royal Soc. London B 263, 1249–1254). Our study attempted to decrease the number of cells required. A sequence within the chromobox-helicase-DNA-binding (CHD) gene, located on the sex chromosomes of all avian species, was amplified. The sequence lengths were 362 and 354 base pairs for the CHD-W and CHD-Z, respectively. The polyacrylamide gel electrophoresis (PAGE) purified primers used were 5´-TCTGCATCGCTAAATCCTTT-3´ and 5´-CTCCCAAGGATGAGRAAYTG-3´ (2.5 µm) (IDT, Inc., San Jose, CA, USA). All assays used Taq DNA polymerase (2.7 U) (M0273L, New England BioLabs, Ipswich, MA, USA) and deoxyribonucleotides (5 µm) (C01581, GenScript Corp, Piscataway, NJ, USA). Lymphocytes from chickens (Gallus domesticus) (10 cells/2 µL) were used as the DNA source for all experiments. Assays were run with positive and negative DNA controls. The DNA was replicated in a Corbett Rapid Thermocycler (Model FTS-IS, Corbett Research, Sydney, Australia) in 20 µL volumes with an annealing temperature of 48°C. All of the PCR products were separated using PAGE. An 8% gel (17:1, con- to bis-acrylamide) with 10 mm TRIS (pH 8) was formed in an agarose gel chamber (M12 Electrophoresis Unit, Edvotek, Bethesda, MD, USA) under Ar. The gel was placed in 10 mm TRIS (pH 8) in the electrophoresis apparatus and the PCR products were added to wells. The applied voltage was 200 and the duration was 2 h (PS500ST, Hoefer Scientific Instruments, San Francisco, CA, USA). The gel was stained for 30 min in 1.25 µm ethidium bromide in 100 mL of 10 mm TRIS (pH 8). Destaining was carried out over 45 min in 100 mL of H2O. The gel was viewed using a transilluminator (3–300, Fotodyne, Hartland, WI, USA) and photographed with an Olympus digital camera. An initial experiment established Griffiths' assay in our lab. Results were consistent with published data, albeit with the same troubling signal-to-noise problems. No signals were observed in assays with less than 20 000 cells. The next experiment compared the use of Griffiths' amplification buffer to a buffer we developed, Bart: 50 mm barbital, 1% dextran T-500, 50 mm KCl, 2.5 mm MgCl2, and 0.035% 2-mercaptoethanol. Signals were produced and a working assay was established with only 10 cells needed, significantly fewer cells than the 20 000 cells necessary for Griffiths' protocol. It was possible to increase the number of productive replication cycles from 35 to 45 without generation of noise. In fact, use of Bart eliminated primer-generated noise, leaving only sexing bands in the gel. Interestingly, when Bart was used with 20 000 cells, no signals were observed. Assays incorporating Bart were run in triplicate and signals were consistently observed. Reduction in the number of cells required for avian sex determination provides potential applications for the sexing of embryos or sexing from a single down feather. Our assay makes sex determination prior to hormonal treatment simple. We are currently replacing the use of the CHD gene with a conserved W-specific sequence.