Statistical methodology has always played a fundamental role in modern animal breeding and genetics. For example, regression and ANOVA techniques have been developed and applied extensively in the context of estimation of genetic parameters and prediction of genetic merit for complex traits. Later, linear mixed model approaches such as best linear unbiased prediction (BLUP) and residual maximum likelihood (REML) estimation of (co-)variance components became prevailing in the analysis of pedigreed data, given their flexibility to accommodate unbalanced data, complex genetic relationships and overlapping generations. Several extensions of mixed models techniques have also been applied in animal breeding, such as the analysis of binary and count data, growth curves, survival models, and gene mapping in outbred populations. These complex models have been frequently implemented using Bayesian and MCMC techniques, facilitated by recent advances in computing technology. More recently, accessibility to genomic technologies has allowed fine mapping of causative loci, high throughput functional genomics studies, and whole-genome prediction of complex traits in livestock species. However, advancements in genomic technologies have also brought several new challenges from data-storage and data-mining standpoints, given the dimensionality of current data sets. Nowadays, not only efficient computer algorithms are required for data storage and data management, but also carefully tailored data mining tools are essential to deal with issues of multiple testing, potential of over fitting, spurious associations, and nonlinearities and complex interactions inherent to genomic data. In this presentation I will review some of the contemporary statistical and data mining methods currently used in animal breeding and genetics, for both prediction and causal inference, with especial emphasis on mixture regression models and graphical models, and the incorporation of biological knowledge into the analyses. Through some examples, I will illustrate the importance of complex modeling in the age of genomic selection.