We here describe two proteomic datasets deposited in Proteo-meXchange via PRIDE partner repository [1] with dataset identifiers PXD000484 (defined as " training ") and PXD000485 (defined as " test ") that have been used for the development of a tamoxifen outcome predictive signature [2]. Both datasets comprised 56 fresh frozen estrogen receptor (ER) positive primary breast tumor specimens derived from patients who received tamoxifen as first line therapy for recurrent disease. Patient groups were defined based on time to progression (TTP) after start of tamoxifen therapy (6 months cutoff): 32 good and 24 poor treatment outcome patients were comprised in the training set, respectively. The test set included 41 good and 15 poor treatment outcome patients. All specimens were subjected to laser capture microdissection (LCM) to enrich for epithelial tumor cells prior to high resolution mass spectrometric (MS) analysis. Protein identification and label-free quantification (LFQ) were performed with Max-Quant software package [3]. A total of 3109 and 4061 proteins were identified and quantified in the training and test set, respectively. We here present the first public proteomic dataset analyzing ER positive recurrent breast cancer by LCM coupled to high resolution MS.