Background: Cancer of unknown primary (CUP) is a common cause of cancer death, with a median survival < 12 months. SUPER is a prospective cohort study designed to create a national information and biobank of patients (pts) with no confirmed primary site following diagnostic work-up. We aimed to determine the impact of tumour molecular profiling on treatment decisions. Methods: 449 pts were recruited (2013-2021) over 3 phases from 12 Australian sites. Clinical information collected over 12 months included: demographics, treatments, investigations and clinico-pathological characteristics. Molecular tests were centrally performed and included comprehensive panel (CCP) sequencing and gene-expression tissue of origin (TOO) assays in Phase 1&2 and CCP and whole genome sequencing (WGS) in Phase 3. The number of genes reported on increased over the 3 phases. Molecular results were discussed in a molecular tumour board. Clinicians completed clinical management questionnaires before and after receiving molecular results. A combined retrospective evaluation of therapeutic actionability was applied to all DNA sequencing data using the TOPOGRAPH database. Results: Median age 63 years [19-86], with 81% ECOG 0-1 and 87% classified as unfavourable CUP subtype. Reporting timelines and rates of successful sequencing improved over time (Table). Clinically actionable genomic abnormalities were detected in 95/331 (29%) pts. Moderate or high match results on TOO assay were consistent with a suspected or later confirmed TOO as reported by clinicians in 81/195 (42%) pts in phase 1 + 2. Germline pathogenic mutations were detected in 34/331 (10%) pts. Molecular tests confirmed clinician’s treatment (started while awaiting results) was consistent with the most likely TOO in 170/331 (51%) pts. Clinicians reported that molecular tests resulted in a change in management in 11/118 pts (9%) in phase 1 and 40/213 (19%) pts in phases 2+3, with 17/51 (33%) pts changed treatment based on potential TOO determination. 11/51 (22%) pts could access treatment via standard pathways given a more specific tumour-type diagnosis and 15/51 (29%) pts were potentially eligible for clinical trials; however, only 7 pts were well enough for referral. Conclusions: The clinical impact of molecular testing in CUP improved over time as testing became more sophisticated and turnaround times improved. Clinicians suspected TOO pre-profiling was consistent with molecular results in half of the pts. Routine access to novel therapies in CUP remains a challenge.