The objective of this study was to predict genomic breeding values for preweaning mortality and birth weight using a bivariate threshold-linear maternal effect model. The data for preweaning mortality and birth weight, and genotypes were obtained from PIC North America and contained 123,163 phenotypic records and 135,530 animals in pedigree. The genotype file contained 42,787 single nucleotide polymorphism (SNP) markers for 13,566 pigs. Birth weight and preweaning mortality were recorded at piglet level. The model included fixed contemporary group, sex, parity, and number of total born per litter effects, random direct and maternal genetic effects, random maternal permanent environmental effects (MPE), and random residual effects (R). First, direct and maternal heritabilities for preweaning mortality and birth weight and genetic correlations were estimated with the Gibbs sampling THRGIBBS1F90 program. Second, genomic breeding values (GEBV) were predicted with the CBLUP90IOD program, which uses preconditioning conjugate gradients and iteration-on-data for a large number of categorical records, using a single-step genomic BLUP. Direct heritability estimates for preweaning mortality and birth weight were 5.4 and 14.0%, respectively. Maternal heritability estimates were 2.1 and 17.2%. Direct and maternal genetic correlations between the two traits were −0.18 and −0.63, respectively. Environmental correlations were also negative (−0.34 and −0.41 for MPE and R). Correlations of GEBV for 13,566 genotyped animals between the two traits were −0.17 and −0.84 for direct and maternal genomic effects, respectively. Preweaning mortality is negatively affected by birth weight especially through maternal genetic components.