Various chemical forms of silicon (Si) possibly present in petroleum products/fuels were investigated in this work: low-molecular-mass species with different compositions, structures, and volatilities, as well as a Conostan oil standard, polysiloxanes, and silicon dioxide. Analysis in xylene solutions was accomplished using wavelength-dispersive X-ray fluorescence (WD XRF) or high-resolution continuum source flame atomic absorption spectrometry (HR-CS FAAS). Samples were also analyzed using HR-CS FAAS in aqueous solutions after mineralization in a closed microwave-heated system, mineralization in an open system, or mineralization combined with fusion using lithium metaborate. It was found that none of the applied procedures was suitable for the accurate determination of Si in all the tested forms. The total content of Si can only be determined if various methodological variants are applied. Using WD XRF, similar signals are obtained for all the investigated organic Si forms. The procedure is also helpful to detect sedimentation (presence) of inorganic Si compounds (successive measurements in the same cup). In direct measurements of organic solutions using HR-CS FAAS, analytical response is similar for all the investigated organic forms, apart from the significantly increased signal for the most volatile forms. On the other hand, underestimated results due to losses of volatile and moderately volatile Si compounds and/or difficulty in analyte transfer to an aqueous solution in the mineralization procedures were stated. The procedure combining mineralization and fusion gave quantitative results for SiO2, the Conostan standard, and some polysiloxanes. The similar behavior of Si compounds belonging to a given group (organic very volatile, moderately volatile, nonvolatile, or inorganic), leading to a systematic error in a given procedure, was used to acquire knowledge on the presence of the kind of Si species in a sample. The proposed analytical scheme employs easy available analytical techniques and can be widely applied. It was also found that internal standard calibration available in HR-CS FAAS does not correct chemical interference due to the LiBO2 matrix. However, using vanadium (V) as the internal standard, it is possible to correct sensitivity drift with time.