Epidemiologic studies have established that pregnancy has a bidirectional, time-dependent effect on breast cancer risk; a period of elevated risk is followed by a long-term period of protection. The purpose of the present study was to determine whether pregnancy and involution are associated with gene expression changes in the normal breast, and whether such changes are transient or persistent. We examined the expression of a customized gene set in normal breast tissue from nulliparous, recently pregnant (0-2 years since pregnancy), and distantly pregnant (5-10 years since pregnancy) age-matched premenopausal women. This gene set included breast cancer biomarkers and genes related to immune/inflammation, extracellular matrix remodeling, angiogenesis, and hormone signaling. Laser capture microdissection and RNA extraction were done from formalin-fixed paraffin-embedded reduction mammoplasty and benign biopsy specimens and analyzed using real-time PCR arrays containing 59 pathway-specific and 5 housekeeping genes. We report 14 of 64 (22%) of the selected gene set to be differentially regulated (at P < 0.05 level) in nulliparous versus parous breast tissues. Based on gene set analysis, inflammation-associated genes were significantly upregulated as a group in both parous groups compared with nulliparous women (P = 0.03). Moreover, parous subjects had significantly reduced expression of estrogen receptor alpha (ERalpha, ESR1), progesterone receptor (PGR), and ERBB2 (Her2/neu) and 2-fold higher estrogen receptor-beta (ESR2) expression compared with nulliparous subjects. These initial data, among the first on gene expression in samples of normal human breast, provide intriguing clues about the mechanisms behind the time-dependent effects of pregnancy on breast cancer risk.