Background: The addition of palbociclib to fulvestrant improved clinical outcomes over placebo-fulvestrant in endocrine-pretreated metastatic breast cancer (MBC) patients in PALOMA-3. Here, we examined factors predictive of long-term benefit. Methods: Premenopausal-peri/postmenopausal patients with endocrine-resistant, hormone receptorepositive (HRþ)/human epidermal growth factor receptor 2enegative MBC were randomised 2:1 to fulvestrant (500 mg) and either palbociclib (125 mg/d; 3/1 schedule; n Z 347) or placebo (n Z 174). Baseline characteristics, mutation status and HR expression levels were compared in patients with and without prolonged benefit (treatment duration !18 months). Results: By August 2016, 100 patients (29%) on palbociclib-fulvestrant and 26 (15%) on placebo-fulvestrant demonstrated prolonged benefit, with long-term responders in both arms sharing common clinical characteristics. They usually had less disease burden at baseline versus those treated <18 months, such as having one disease site (40% vs 29% on palbociclib-fulvestrant and 69% vs 29% on placebo-fulvestrant), bone-only disease (32% vs 22% and 46% vs 17%) and were less heavily pretreated (69% vs 56% and 73% vs 60% had 2 prior therapies). Baseline tumour ESR1 and PIK3CA mutation rates were lower among long-term responders in both arms; median oestrogen receptor H-scores were similar, whereas progesterone receptor H-scores were higher among long-term responders. Conclusions: This exploratory analysis demonstrates that some patients with endocrineresistant MBC derive significant and prolonged benefit when treated with palbociclibfulvestrant, with fewer patients experiencing similar efficacy with placebo-fulvestrant. The current analysis did not identify specific molecular or clinical factors prognostic of long-term benefit with palbociclib-fulvestrant (ClinicalTrials.gov, NCT01942135).