There have been numerous attempts to determine the channel capacity of a nonlinear fiber optics communication channel. The main approach, until recently, was to consider ASE noise as a predominant effect and to observe the fiber nonlinearities as the perturbation of linear case or as the multiplicative noise. In this invited paper, we describe how to determine the true fiber-optics channel capacity. Because in most of practical applications the channel input distribution is uniform, we also describe how to determine the uniform information capacity, which represents the lower bound on channel capacity. This method consists of two steps: 1) approximating probability density functions (PDFs) for energy of pulses, which is done by one of the following approaches: (a) evaluation of histograms, (b) instanton approach or (c) edgeworth expansion, and 2) estimating information capacities by applying a method originally proposed by Arnold and Pfitser.