The likelihood for copula modeling appears when both the data and the copula representations are seen as being driven by common uniform latent variables. This perspective facilitates Bayesian inference for prediction and copula selection.