Sex allocation studies seek to ascertain whether mothers manipulate offspring sex ratio prior to ovulation. To do so, DNA for molecular sexing should be collected as soon after conception as possible, but instead neonates are usually sampled. Here, we aim to identify and quantify some of the problems associated with using molecular techniques to identify the sex of newly laid avian eggs. From both fertilized and unfertilized chicken (Gallus gallus) eggs, we sampled (1) the blastoderm/disc, (2) vitelline membrane and (3) a mixture of (1) and (2). Thus, we replicated scenarios under which contaminated samples are taken and/or unfertilized eggs are not identified as such and are sampled. We found that two commonly used molecular sexing tests, based on the CHD-1 genes, differed in sensitivity, but this did not always predict their ability to sex egg samples. The vitelline membrane was a considerable source of maternal and probably paternal contamination. Fertile eggs were regularly assigned the wrong sex when vitelline membrane contaminated the blastoderm sample. The membrane of unfertilized eggs was always female, i.e. maternal DNA had been amplified. DNA was amplified from 47 to 63% of unfertilized blastodiscs, even though it was highly unlikely that DNA from a single haploid cell could be amplified reliably using these polymerase chain reaction (PCR) techniques. Surprisingly, the blastodiscs were identified as both males and females. We suggest that in these cases only maternal DNA was amplified, and that ‘false’ males, Z not ZZ, were detected. This was due to the reduced ability of both sets of primers to anneal to the W chromosome compared to the Z chromosome at low DNA concentrations. Overall, our data suggested that estimates of primary sex ratios based on newly laid eggs will be appreciably inaccurate.