A Bayesian framework is developed to quantify predictive uncertainty in environmental modeling caused by uncertainty in modeling scenarios, model structures, model parameters, and data. An example of using the framework to quantify model uncertainty is presented to simulate soil microbial respiration pulses in response to episodic rainfall pulses (the “Birch effect”). A total of five models are developed; they evolve from an existing four-carbon (C) pool model to models with additional C pools and recently developed models with explicit representations of soil moisture controls on C degradation and microbial uptake rates. Markov chain Monte Carlo (MCMC) methods with generalized likelihood function (not Gaussian) are used to estimate posterior parameter distributions of the models, and the posterior parameter samples are used to evaluate probabilities of the models. The models with explicit representations of soil moisture controls outperform the other models. The models with additional C pools for accumulation of degraded C in the dry zone of the soil pore space result in a higher probability of reproducing the observed Birch pulses. A cross-validation is conducted to explore predictive performance of model averaging and of individual models. The Bayesian framework is mathematically general and can be applied to a wide range of environmental problems.