This work presents a statistical model for the propagation of noise through interferometric demodulation processes. Using the case of a 3×3 passive digital demodulation algorithm, an exact transfer function and its associated probability structure for intensity noise propagation through a fiber Bragg grating sensor interrogation system is derived. This new model is generalized to any input noise probability structure and includes the possibility of full or partial correlation among the demodulation input channels. This work then presents results for the specific case of Gaussian intensity noise (with and without channel correlations) and shows explicit interferometer phase influence on output noise statistical moments, which are important for signal-to-noise predictions. The demodulators nonlinear transfer function is shown to induce output bias as well as either attenuate or amplify output variance, depending upon the signal phase. Experimental data are provided to validate the model. This model generalizes to support predicting output noise levels in Bragg grating-based sensing systems.© (2011) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.