The rapid evolution of larger and more detailed data sets in genetics requires new statistical modeling tools and inferences that are flexible to different kinds of data and are also fast enough to enable the analyses. Hierarchical generalized linear models (HGLMs) based on the h-likelihood is such a tool. A single algorithm, iterative weighted least squares, can be used and requires neither prior distributions of parameters nor multi-dimensional quadrature. HGLMs combine modeling flexibility, previously only possible with Bayesian methods, with simple regression based algorithms. The theory behind HGLM and the computational aspects of it will be presented together with some examples from genetics. I have recently developed an R package "hglm" available on CRAN and is easily downloaded in R.