A collection of C++ classes for Bayesian modeling, with abstractions for Model, Parameter, Data, and Method. Also includes lots of R-like supporting material including wrappers for LAPACK/BLAS linear algebra, random number generation, numerical optimization, integration, and MCMC samplers. Implemented models include many generalized linear models, most exponential families, finite mixture models, hidden Markov models, and some basic dynamic linear models. BOOM is designed to be an extensible back end for Bayesian computing, so implementing other models should be straightforward.