Genome-wide cfDNA fragmentation patterns have previously been demonstrated to distinguish with high sensitivity and specificity between plasma samples from individuals with and without cancer. To further evaluate cfDNA fragmentation as a blood-based screening test for cancer, we have used low coverage whole genome sequencing to analyze plasma samples from 280 patients referred to an advanced diagnostic center due to non-organ specific signs and symptoms of cancer. Within three months of inclusion, 74 of these patients were diagnosed with one of 16 different solid cancers while 206 patients did not have cancer. Using an improved version of our genome-wide cfDNA fragmentation analyses, we observed high performance in distinguishing cancer and non-cancer samples (AUC=0.92, 95% CI 0.88-0.96), including lung cancer (n=12, AUC=0.91, 95% CI 0.80-1.00) and colorectal cancer (n=12, AUC=0.94, 95% CI 0.89-0.99). Although many of the patients in this cohort had other common illnesses including cardiovascular, autoimmune, and inflammatory diseases, the machine learning models of cfDNA fragmentation were able to detect cancer with high sensitivity and specificity. These data support the development of genome-wide cfDNA fragmentation analyses as a non-invasive detection screening approach for both single and multiple cancers. Citation Format: Jacob Carey, Alessandro Leal, Bryan Chesnick, Denise Butler, Michael Rongione, Sian Jones, Rob Scharpf, Mette Villadsen, Stig E. Bojesen, Julia S. Johansen, Claus L. Feltoft, Victor E. Velculescu, Nicholas C. Dracopoli. Detecting cancer using genome-wide cfDNA nucleosomal fragmentation in a prospective multi cancer cohort [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 570.