In recent years, several environmental pollutants have been monitored in surface waters and sediments. However, few studies apply multivariate statistics to identify the main components and correlate them temporally and spatially. In this sense, the present study sought to monitor the quality of water and sediments in the Rio Marrecas/Brazil, through the analysis of physicochemical parameters and trace elements, as well as to identifying sources of contamination, using multivariate statistics. For this purpose, sampling was carried out in nine locations for a period of 12 months. The Total Reflection X-ray Fluorescence (TXRF) technique was used to quantify the 15 elements identified in water and sediment samples. Through multivariate statistical analyses, the most significant elements, their correlations and possible pollutant sources were defined, and the pollution index (HPI) and assessment index (HEI) of heavy metals were applied. The parameters pH and BOD5 do not comply with Brazilian legislation. Based on PCA and Spearman correlation, there was strong evidence of contamination of the water naturally, composed of the elements Ti, V, Mn, Fe, and of anthropogenic origin composed of the elements Ca, Ni, Cu, Zn. These findings provide insights to determine the impacts of heavy metals on human health and the environment.