Haplotype-based techniques are being used to estimate the relative age of alleles--particularly in screening loci for signals of recent positive selection--but does this approach capture even coarse age differences? Using simulations and empirical data from the International HapMap Project, we show that a simple pairwise metric of haplotype homozygosity gives significantly higher mean values for human single-nucleotide-polymorphism alleles that appear to be derived than for those that appear to be ancestral, as determined by comparison with the chimpanzee genome. Our results support the use of haplotype-based techniques, such as extended haplotypic homozygosity, to assess the age of alleles.