Immunotherapy is an important and established treatment option for patients with advanced melanoma. Initial anti-PD1 trials arbitrarily defined a two-year treatment duration, but a shorter treatment duration may be appropriate. In this study, we retrospectively assessed 70 patients who stopped anti-PD1 therapy in the absence of progressive disease (PD) to determine clinical outcomes. In our cohort, the median time on treatment was 11.8 months. Complete response was attained at time of anti-PD1 discontinuation in 61 (87%). After a median follow up of 34.2 months (range: 2–70.8) post discontinuation, 81% remained disease free. Using ddPCR, we determine the utility of circulating tumour DNA (ctDNA) to predict progressive disease after cessation (n = 38). There was a significant association between presence of ctDNA at cessation and disease progression (p = 0.012, Fisher’s exact test) and this conferred a negative and positive predictive value of 0.82 (95% CI: 0.645–0.930) and 0.80 (95% CI 0.284–0.995), respectively. Additionally, dichotomised treatment-free survival in patients with or without ctDNA at cessation was significantly longer in the latter group (p < 0.001, HR: 0.008, 95% CI: 0.001–0.079). Overall, our study confirms that durable disease control can be achieved with cessation of therapy in the absence of disease progression and undetectable ctDNA at cessation was associated with longer treatment-free survival.