Relative survival is the ratio of overall survival (OS) over survival of the general population, and widely used in epidemiological studies. But it is artificially higher than OS and thus inferior to OS for cancer prognostication of individual patients. Moreover, trend-changes and disparities in OS of breast cancer are unclear while the relative survival of breast cancer has been reported on a regular basis. Therefore, we estimated trends in age-standardized 5-year OS of invasive breast cancer, using data from the Surveillance, Epidemiology, and End Results (SEER) cancer registry program and piecewise-linear regression models. Among 188,052 women with breast cancer diagnosed during 2007-2010 (SEER-18, 155,515 [79.3%] survived by year 5), the 5-year OS significantly differed by age, histology, tumor grade, tumor stage, hormone receptors, race/ethnicity, insurance status, region, rural-urban continuum and selected county-attributes. Among 469,498 women with breast cancer diagnosed during 1975-2010 (SEER-9) in the U.S., we observed an upward trend in the age-standardized 5-year OS (stage- and race/ethnicity-adjusted annual percentage change = 0.97 [95% CI, 0.76-1.18]). The 36-year trends/slopes in age-standardized 5-year OS of breast cancer differed by histology, tumor grade, stage, race/ethnicity, region and socioeconomic attributes of the patient's residence-county, but not by those of rural-urban continuum. The 3-joinpoint model on the 36-year trend identified significant slope changes in 1983, 1987 and 2000, with the largest slope (2.5%/year) during 1983-1987. In conclusion, we here show trends in the age-standardized 5-year OS among U.S. women with breast cancer changed in diagnosis-years of 1983, 1987 and 2000, and differed by tumor characteristics and race/ethnicity. More efforts are needed to understand the trend changes and to address the OS disparities of breast cancers.