We propose a novel approximate-likelihood method to fit demographic models to human genomewide single-nucleotide polymorphism (SNP) data. We divide the genome into windows of constant genetic map width and then tabulate the number of distinct haplotypes and the frequency of the most common haplotype for each window. We summarize the data by the genomewide joint distribution of these two statistics-termed the HCN statistic. Coalescent simulations are used to generate the expected HCN statistic for different demographic parameters. The HCN statistic provides additional information for disentangling complex demography beyond statistics based on single-SNP frequencies. Application of our method to simulated data shows it can reliably infer parameters from growth and bottleneck models, even in the presence of recombination hotspots when properly modeled. We also examined how practical problems with genomewide data sets, such as errors in the genetic map, haplotype phase uncertainty, and SNP ascertainment bias, affect our method. Several modifications of our method served to make it robust to these problems. We have applied our method to data collected by Perlegen Sciences and find evidence for a severe population size reduction in northwestern Europe starting 32,500-47,500 years ago.A major goal of evolutionary genetics is to infer the demographic history of a population. This is traditionally done by fitting a population genetic model to sequence data taken from a sample of individuals. The population genetic model often includes parameters allowing for changes in population size or population structure with or without migration. Such parameters are interesting in their own right, but are critical to define a proper ''null model'' that can be used to find ''unusual'' genes that may be targets of positive or negative selection (Jensen et al. 2005). Additionally, a proper demographic model is important for assessing genomewide patterns of positive and negative selection Lohmueller et al. 2008).Methods have been developed that make full use of sequence data to infer demographic parameters (Griffiths and Tavaré 1994;Kuhner et al. 1995). These methods are computationally intensive and are impractical for all but the smallest data sets. Thus, researchers have turned to methods based on summary statistics (reviewed in Marjoram and Tavaré 2006). Summary statistics can be quickly calculated from the data and then be used to infer model parameters using either a likelihood or approximate Bayesian computation (ABC) framework (for example, Wall 2000a; Fagundes et al. 2007). The key for successful application of this approach is to find summaries of the data that contain enough information about the demographic parameters of interest. One of the most successfully used summary statistics for population genetic inference, the site frequency spectrum (SFS) (Nielsen 2000;Adams and Hudson 2004;Caicedo et al. 2007;Hernandez et al. 2007b), is a sufficient statistic for the full data if the singlenucleotide polymorphisms (SNPs) ...