Bone weight, defined as the total weight of the bones in all the forequarter and hindquarter joints, can reflect somebody conformation traits and skeletal diseases. To gain a better understanding of the genetic determinants of bone weight, we used a composite strategy including multimarker and rare-marker association to perform genomewide association studies (GWAS) for that character in Simmental cattle. Our strategy consisted of three models: (i) A traditional linear mixed model (LMM) was applied (Q+K-LMM); (ii) single nucleotide polymorphisms (SNPs) with p-values less than .05 from the LMM were selected to undergo the least absolute shrinkage and selector operator (Lasso) in the second stage (LMM-Lasso); (iii) genes containing two or more rare SNPs were examined by performing the sequence kernel association test (gene-based SKAT). A total of 1,225 cattle were genotyped with an Illumina BovineHD BeadChip containing 770,000 SNPs. After the quality-control procedures, 1,217 individuals with 608,696 common SNPs and 105,787 rare SNPs (with 0.001 < minor allele frequency [MAF] <0.05) remained in the sample for analysis. A traditional LMM successfully mapped three genes associated with bone weight, while LMM-Lasso identified nine genes, which included all genes found by traditional LMM. Only a single gene, EPHB3, surpassed the significance threshold after Bonferroni correction in gene-based SKAT. In conclusion, based on functional annotation and results from previous endeavours, we believe that LCORL, RIMS2, LAP3, PRKAR2B, CHSY1, MAP2K6 and EPHB3 are candidate genes for bone weight. In general, such a comprehensive strategy for GWAS may be useful for researchers seeking to probe the full genetic architecture underlying economic traits in livestock.