With the increasing world population and the consequent increasing in industrial activities and fuel consumption, chemical pollution has been a major problem of the XXI century. Alternatives that can minimize such problems are targets of researches, such as the use of biodiesel, a renewable and less polluting fuel. Elements usually found in low concentrations, such as As, Sb and Se, become toxic to human health when in concentrations above certain limits and, therefore, should be monitored in several samples, such as fuels. For Sb the acceptable daily intake (ADI) is 6 mg day(-1) for each kg of body weight and, for Se, concentrations above 50-200 mg per day are considered toxic. The As acceptable limit in drinking water is 0.01 mg L(-1) and values above 7.5 x 10-3 g m(-3) are associated to risk of lung cancer. These elements may be original constituents of petroleum and its derivatives or be added as contaminants at some step of the refining process. In the case of biodiesel, these elements can be present due to the use of fertilizers and pesticides in the crops that produce the oil used in the biodiesel production. These elements can be released to the atmosphere during the fuel burning in the motor. The usually low concentrations of these elements can be below the detection limit of some techniques, such as ICP OES with sample introduction by standard pneumatic nebulization. In these case, the vapor generation is an interesting alternative for sample introduction, since it provides a significant improvement in the sensitivity of the analytical technique. This study aimed to simultaneously determine As, Sb and Se in crude oil and biodiesel samples, using HG-ICP OES. For that, samples had to be subjected to acid decomposition in a block digester, thus allowing calibration with inorganic standard solutions. Different pre-reduction conditions were studied (HCl, thiourea and ascorbic acid with KI), being evaluated the reagents concentrations and heating time. The conditions using HCl and thiourea showed the best results and had their parameters multivariate optimized, as well as the parameters of the vapor generation: sample, sodium borohydride and carrier gas flow rates. Established the best conditions for analysis and respecting the limitations of the technique, the method was validated by analysis of a residual fuel oil certified material (NIST 1634c), biodiesel from soybean oil (NIST 2772) and biodiesel from animal fat (NIST 2773), using the two cited pre-reduction conditions. In this case, the use of HCl as pre-reducer was more efficient, with recoveries between 90% and 99% for the three analytes. Pre-reduction with thiourea was effective only for Sb, with recovery around 104%, while the recoveries for As and Se were between 57% and 29%. Thus, HCl was employed as the pre-reducer in the proposed method. The optimized conditions for the generation and used for the analysis of the samples were: sample flow rate, 3.0 mL min(-1); NaBH4 flow rate, 1.5 mL min(-1); carrier gas flow rate, 0.8 L min(-1) and concentration of HCl, 8 mol L(-1). A study to verify a possible interference of Ni under the optimized conditions for vapor generation revealed that this interference does not affect the determination of As, Sb and Se when employing HCl 8 mol L(-1). Therefore, crude oil and biodiesel samples were analyzed at the previously optimized conditions. Only As and Se were detected in crude oil samples, while the concentrations of these analytes in the biodiesel samples were always lower than the obtained LODs. Although, the recovery for the Sb additions in biodiesel and crude oil samples were in a large range, between 72% and 111%, indicating that HCl is not the most efficient pre-reducer for this element, it can be said that, for multielemental analysis whose objective is to evaluate the presence of this element or a possible contamination by it, this prereducer can be employed.