Nrf2 (Nuclear factor E2 p45-related factor 2) is a transcription factor that activates the transcription of many cytoprotective genes to combat oxidative and electrophilic stresses. This regulation is accomplished by the binding of Nrf2 to a DNA segment called antioxidant response element (ARE) located in the vicinity of target genes. Elucidating these direct targets of Nrf2 can provide potential therapeutic insights against oxidative diseases, including cancer. In this project, we strive to improve the current Nrf2 binding prediction model. Using genome-wide chromatin immunoprecipitation followed by parallel sequencing (ChIP-Seq), we identified ~1200 bound DNA sequences representing a 50-fold increase in the number of known Nrf2 sites. A motif discovery software, MEME, identified a 11 bp motif in the Nrf2 ChIP-Seq dataset, consistent with the current Nrf2 model. High scoring Nrf2 sites cluster in greater number compared to low scoring Nrf2 sites and background. We further identified dinucleotide dependencies AA or TT 2bp downstream of the Nrf2 sites. DNA bendability analysis with Emboss BANANA reveals a decrease in DNA bendability at those dependent positions, consistent with properties of A and T nucleotides. Incorporating these newly identified properties results in dramatic increase in specificity, but also leads to reduced sensitivity. A receiver operating characteristic (ROC) analysis illustrates the need for implementing a more sophisticated probabilistic model to better capture the detected properties.