Abstract Background Continuing medical education (CME) in earlier cancer diagnosis was launched in Denmark in 2012 as part of the Danish National Cancer Plan. The CME programme was introduced to improve the recognition among general practitioners (GPs) of symptoms suggestive of cancer and improve the selection of patients requiring urgent investigation. This study aims to explore the effect of CME on GP knowledge about cancer diagnosis, attitude towards own role in cancer detection, self-assessed readiness to investigate and cancer risk assessment of urgently referred patients. Methods We conducted a before-after study in the Central Denmark Region including 831 GPs assigned to one of eight geographical clusters. All GPs were invited to participate in the CME at three-week intervals between clusters. A questionnaire focusing on knowledge, attitude and clinical vignettes was sent to each GP one month before and seven months after the CME. The GPs were also asked to assess the risk of cancer in patients urgently referred to a fast-track cancer pathway during an eight-month period. CME-participating GPs were compared with reference (non-participating) GPs by analysing before-after differences. Results One quarter of all GPs participated in the CME. 202 GPs (24.3 %) completed both the baseline and the follow-up questionnaires. 532 GPs (64.0 %) assessed the risk of cancer before the CME and 524 GPs (63.1 %) assessed the risk of cancer after the CME in urgently referred consecutive patients. Compared to the reference group, CME-participating GPs statistically significantly improved their understanding of a rational probability of diagnosing cancer among patients urgently referred for suspected cancer, increased their knowledge of cancer likelihood in a 50-year-old referred patient and lowered the assessed risk of cancer in urgently referred patients. Conclusions The standardised CME lowered the GP-assessed cancer risk of urgently referred patients, whereas the effect on knowledge about cancer diagnosis and attitude towards own role in cancer detection was limited. No effect was found on the GPs’ readiness to investigate. CME may be effective for optimising the interpretation of cancer symptoms and thereby improve the selection of patients for urgent cancer referral. Trial registration NCT02069470 on ClinicalTrials.gov. Retrospectively registered, 1/29/2014.