The ancestral chromosomal segments in admixed genomes are of significant importance for both population history inference and admixture mapping, because they essentially provide the basic information for tracking genetic events. However, the distributions of the lengths of ancestral chromosomal segments (LACS) under some admixture models remain poorly understood. Here we introduced a theoretical framework on the distribution of LACS in two representative admixture models, that is, hybrid isolation (HI) model and gradual admixture (GA) model. Although the distribution of LACS in the GA model differs from that in the HI model, we demonstrated that the mean LACS in the HI model is approximately half of that in the GA model if both admixture proportion and admixture time in the two models are identical. We showed that the theoretical framework greatly facilitated the inference and understanding of population admixture history by analyzing African-American and Mexican empirical data. In addition, we found the peak of association signatures in the HI model was much narrower and sharper than that in the GA model, indicating that the identification of putative causal allele in the HI model is more efficient than that in the GA model. Thus admixture mapping with case-only data would be a reasonable and economical choice in the HI model due to the weak background noise. However, according to our previous studies, many populations are likely to be gradually admixed and have pretty high background linkage disequilibrium. Therefore, we suggest using a case-control approach rather than a case-only approach to conduct admixture mapping to retain the statistics power in recently admixed populations.