The Psunj Mt. is the highest mountain in the Slavonia region of Croatia. Peaking at 985 m above sea level, it is elliptical in shape with streams cut through ridges that extend radially from its central part. The mountain is covered with forest, except inhabited marginal parts. Geologically, its crystalline core is composed of various types of metamorphic and igneous rocks. Metamorphic rocks are represented mostly by gneisses, schists, amphibolites and flaser granitoids. Some of these rocks are Precambrian (questionable) in age and form the basement rocks of the Pannonian Basin. The latest recorded metamorphic event occurred during the Alpine orogeny in the Upper Cretaceous. Igneous rocks are represented mostly by plutonic rocks of various granitic compositions, intruded during the Hercynian orogeny in the Devonian. Siliciclastic sedimentary rocks formerly covered the entire area of the Psunj Mt., but since its uplift during the Miocene, they have been partially eroded and today exist mainly in marginal and central parts. All rocks at the Earth's surface are exposed to the processes of weathering. During weathering processes rocks break down, releasing chemical elements and minerals in streams. Unstable elements and minerals form more stable complexes and secondary minerals. Both, old and new minerals and complexes are deposited in favourable conditions, and in addition to organic matter, make up the stream sediment. A stream sediment sample collected at some point represents the whole drainage basin, that is, the area upstream from that point bounded by basin divide. Because of its representativeness, the stream sediment can be useful sampling medium in geochemical studies. Within the conducted geochemical study of the Psunj Mt., 361 stream sediment samples were collected and their chemical composition was determined. The concentrations of analysed elements represent the basic input data. Such data are compositional, that is, mutually dependent, and require the use of appropriate multivariate statistical methods for processing. Accordingly, the statistical methods used in this study were based on the relationships among the concentrations of elements and not on their absolute values. Although investigations of stream sediment have been carried out in the studied area in the past, none of the studies were carried out on such large scale and number of samples so far, nor have the data been processed by statistical methods designed for compositional data. Given that the study area is lithologically diverse, mostly the mixed stream sediment was expected. Therefore, the following questions arose: 1) Can trends be identified in chemical composition of the mixed sediment, or, in other words, can even small changes in majority of samples produce statistically significant trends in chemical composition? 2) Do multivariate anomalous concentrations of certain elements hold meaningful information and do they point to the origin of anomaly in terms of the weathered rocks? The first aim of the research was to define geochemical trends in chemical compositions of stream sediment samples. The second aim was to determine multivariate anomalous concentrations and their origin. The third aim was to interpret trends and anomalies with respect to their geogenic or anthropogenic origin and other measured variables. The hypothesis of the research was that the statistically significant trends in chemical compositions exist and multivariate anomalous concentrations do hold information about their origin. Another hypothesis was that these trends and anomalies are geogenic, that is, the result of lithologic diversity of the investigated area. With the aim to define geochemical trends, samples were first divided into groups. For this purpose, principal component analysis was performed based on concentrations of major elements: Al, Fe, Mg, Ca, Ti, Na, and K. First two components count for most of the variability and each of them divides samples in two distinct groups that can be described by weathering processes of different types of rocks. Next, regression analysis was performed to define the transition trends that include all four groups of samples, based on the three most variable major elements. In order to interpret the trends, representative samples from each group were further analysed for carbonate, organic matter, and iron (oxy)hydroxide content, as well as for granulometric and semiquantitative mineral composition and cation exchange capacity. In order to find multivariate anomalous samples, the first step was to perform principal component analysis based on concentrations of the most variable elements typical for rocks of felsic and mafic composition, respectively. Indeed, grouping of elements on negative and positive pole of the first principal component axis, respectively, as well as the most of the variabilty explained by that component, showed that the weathering of such rocks represents dominant process affecting the chemical composition of stream sediment. Therefore, in the second step, multivariate anomalous samples were sought in the association of elements characteristic for felsic (Pb, La, Th, Nb, and K), and mafic rocks (Cu, Ni, Co, and Cr). In order to determine the origin of anomalous concentrations, rocks were sampled in drainage basins that correspond to anomalous samples. These rocks were microscopically analysed and then the selected rock samples were analysed on chemical composition. The rock samples were selected in order to represent all sampled varieties of rocks, and selection was based on high content of opaque minerals. Darker grains were extracted from rock samples with the highest amount of opaque minerals and examined by an electron microscope. In order to interpret the anomalies, five anomalous samples were subjected to the same set of additional analyses as the representative samples of groups. An additional contribution to the study of anomalies was achieved by the multivariate analysis of anomalous uranium and thorium concentrations, supplemented with the analysis of radionuclides in stream sediment and water in selected drainage basins. An original scientific contribution of this work lies in definition of statistically significant models of trends in chemical composition of mixed stream sediment material. In addition, multivariate anomalous concentrations proved to be valuable pathfinder in pointing out chemically distinct rocks of the investigated area. Also, the scientific contribution is reflected in the interpretation of trends and anomalies with respect to their origin. These findings are important for understanding the impact of weathering on chemical composition of stream sediment. Finally, they contribute to the geological and geochemical comprehension of rocks on the Psunj Mt., including the basement rocks of the Pannonian Basin as well.