A Dimensão Fractal (DF) representa o nível de complexidade e de ocupação do espaço por um objeto fractal. O objetivo do estudo foi identificar se existe correlação entre os valores de DF obtidos em imagens advindas de do uso de Raios-X em resoluções distintas: a Tomografia Computadorizada de Feixe Cônico (TCFC), a Microtomografia Computadorizada (µCT) e a Radiografia Periapical Digital (RPD). Foi realizado um estudo prospectivo transversal que incluiu pacientes saudáveis maiores de 18 anos, que necessitaram de manipulação óssea para exodontia dos terceiros molares e que possuíam o exame de TCFC prévio à cirurgia. Foi realizada coleta de fragmento ósseo na região retromolar, que foram radiografados no aparelho de RPD. Posteriormente, foram processados e incluídos em metilmetacrilato, então escaneados na µCT. Regiões de Interesse (ROIs) foram criadas separadamente nas porções cortical e trabecular para avaliação da DF: 1) TCFC: 6 ROIs (25 x 25 pixels) em três cortes parassagitais; 2) RDP: 2 ROIs (64 x 64 pixels); 3) µCT: 6 ROIs (90 x 90 pixels). Os valores de DF foram obtidos por dois métodos de processamento de imagem: por meio da binarização (DFc) e pelo método baseado na análise das imagens diretamente em escala de cinza (DFg), utilizando o software ImageJ. Nestas ROIs, além da DF, também foram avaliados dados do histograma de desvio-padrão e média dos valores de cinza para RDP e TCFC, e, para µCT, foram avaliados parâmetros microarquiteturais da cortical óssea e da porção trabecular. Foi realizado o teste de normalidade Shapiro-Wilk, seguido do teste de correlação de Spearman ou Pearson (nível de significância de 5%). Houve correlação negativa entre DFc e DFg na porção cortical na TCFC (p 0,01, r0,61). Na µCT, não houve correlação entre DF e os parâmetros analisados na porção trabecular, e na cortical houve uma tendência de correlação negativa de DFg com a superfície óssea (p 0,05, r-0,48). Entre os parâmetros microarquiteturais, houve correlação significante entre volume ósseo (VO), sendo negativa com superfície óssea (SO) e superfície trabecular (Tb.Sp), positiva com espessura trabecular (Tb.Th) para porção trabecular; para cortical, houve correlação forte e positiva da porosidade com SO e números de poros (P.n), e moderada e negativa com a densidade mineral óssea (DMO). Na RPD não houve correlação entre DF e os dados do histograma na trabecular, e houve apenas na DFg com o desvio-padrão dos valores dos pixels (p 0,004, r-0,66). Na TCFC na porção trabecular a DFc correlacionou com a média do valor dos pixels (p 0,03, r0,74) e a DFg correlacionou com o desvio-padrão dos valores dos pixels (p 0,03, r0,73). Conclui-se que não houve correlação entre os valores de DF nas três modalidades analisadas. A DF analisada nas imagens em escala de cinza correlaciona-se com maior amplitude na variação dos valores dos pixels na porção trabecular, tanto para RPD quanto para TCFC. Os valores de DF obtidos pela binarização tiveram correlação com a média do valor dos pixels para RPD, também somente na porção trabecular. Fractal Dimension (FD) stands for the complexity level and occupation rate of a space by a fractal object. The aim of this study was to identify whether there is a correlation between the FD values obtained in X-ray images with different resolutions: Cone-Beam Computed Tomography (CBCT), Microtomography (μCT), and Digital Periapical Radiography (DPR). A prospective study was conducted including healthy individuals older than 18 years, who required bone manipulation for third molar extraction, and who had a CBCT exam prior to the surgery. A bone fragment was collected in the retromolar region, which was radiographed in digital Xray equipment. Subsequently, the fragments were processed and embedded in methyl methacrylate, then scanned in μCT. Regions of Interest (ROIs) were created separately in the cortical and trabecular parts for evaluation: 1) CBCT: 6 ROIs (25 x 25 pixels) in three parasagittal sections; 2) DPR: 2 ROIs (64 x 64 pixels); 3) μCT: 6 ROIs (90 x 90 pixels). Two processing methods were assessed for FD calculation: one through binarization (FDc) and the other through a method based on the use of grayscale images (FDg), using the ImageJ software. In these ROIs, in addition to the FD, data from the histogram regarding the standard deviation and mean values of the grayscale values for DPR and CBCT were also evaluated, and for μCT, microarchitectural parameters of the cortical bone and trabecular bone were assessed. Shapiro-Wilk normality test was performed, followed by Spearman or Pearson correlation tests (significance level of 5%). There was a negative correlation between FDc and FDg in the cortical part in the CBCT (p 0.01, r-0.61). Regarding μCT results, there was no correlation between FD and the parameters analyzed in the trabecular part, and in the cortical, there was a tendency for a negative correlation of FDg with the bone surface (p 0.05, r-0.48). Among the microarchitectural parameters, there was a significant correlation between bone volume (BV), being negative with bone surface (BS) and trabecular surface (Tb.Sp), and positive with the trabecular thickness (Tb.Th) for trabecular portion, and for cortical there was a strong and positive correlation of porosity with OS and pore numbers (P.n), and moderate and negative correlation with bone mineral density (BMD). In the DPR there was no correlation between FD and histogram data in the trabecular, and there was only in the DFg with the standard deviation of the pixel values (p 0.004, r-0.66). Among the microarchitectural parameters, there was a significant correlation between bone volume (BO), being negative with the bone surface (BS) and trabecular surface (Tb.Sp), and positive with the trabecular thickness (Tb.Th) for the trabecular portion; and for cortical there was a strong and positive correlation of porosity with OS and pore numbers (P.n), and moderate and negative correlation with bone mineral density (BMD). In the DPR there was no correlation between FD and histogram data in the trabecular, and there was only in the DFg with the standard deviation of the pixel values (p 0.004, r-0.66). In the CBCT in the trabecular part, FDc correlated with the mean pixel value (p 0.03, r0.74), and the FDg correlated with the standard deviation of the pixel values (p 0.03, r0.73). In conclusion, there was no correlation between the FD values in the three modalities evaluated. FDc value was related to the mean pixel’s value, while the FDg was related to the variation range of the pixel values in the grayscale.