We consider genetic association analysis that combines data from case-parent trios/sibships and unrelated controls. A general and simple methodology is proposed, using a weighted least-squares approach to combine separate information from the case-parent/case-sibling analysis and the case-unrelated control analysis. The new proposal improves over the existing methods in that it requires no assumptions and estimation on the mating-type distribution. Simulation results show that the new method competes well with the likelihood-based method when the latter is applicable, and achieves substantial power gains over separate analyses in general. Therefore, the proposed combined association analysis can enjoy wide applications, including the multiallele/locus, haplotype, and genome-wide association studies