The central problem in laser sensing of natural organic complexes (NOC) in water, is their identification and determination of their state. This is an essential condition for quantitative characterization of an object by optical methods (e.g. fluorimetry). It is very difficult to solve the NOC identification problem dealing only with spectra. It is necessary to penetrate to the molecular level, and to supplement spectral data with molecular photophysical parameters (absorption and fluorescence cross sections, rates of intermolecular transitions and of intermolecular excitation-energy transfer, etc.). Furthermore, it is necessary to measure these parameters in vivo and in situ, under conditions of absence of accurate a priori data. This can be done only by non-linear laser fluorimetry. In this paper, the results of computer experiments and real experiments, illustrating capabilities of non-linear laser fluorimetry in diagnostics of NOC, are presented. To solve the inverse problem, the method of artificial neural networks was used. it is shown that it is possible to achieve the determination precision of the photophysical parameters not worse than the measurement precision of the saturation curve, using a few-parametric model of the formation process of the NOC fluorescence response at pulse laser excitation, and taking into account NOC specifics.