We present a mathematical model that describes the spatial distribution of photons in fluorescence endoscopic images, resulting in expressions for image signal-to-noise ratio and resolution. This model was applied to quantitative analysis of fluorescence images collected from human colonic mucosa with a fiber-optic and an electronic endoscope. It provides a tool for the design of fluorescence endoscopic imaging systems and for extraction of quantitative information about image features. The results apply generally to endoscopic imaging of remote structures in biological and industrial settings, in which light of weak intensity such as fluorescence as well as reflected white light is used.