Retinoids constitute a very promising class of agents for the chemoprevention or treatment of breast cancer. These retinoids exert their biological activity through two distinct classes of retinoic acid (RA) receptors (R), the RAR isotypes (alpha, beta, and gamma) and the three RXR isotypes (alpha, beta, and gamma) and their numerous isoforms which bind as RXR/RAR heterodimers to the polymorphic cis-acting response elements of RA target genes. With respect to these numerous receptor sub-types, the retinoid-induced effects at the biological level include marked modifications with respect to both cell proliferation and cell death (apoptosis), and also in the induction of differentiation processes. The present study aims to characterize the effect which four retinoids (TTNPB, 9-cis-RA, LGD 1069, 4-HPR) with distinct RAR/RXR binding properties induced on various in vitro and in vivo mouse and human breast cancer models. The experiments with the retinoids were carried out in comparison with the anti-estrogen tamoxifen and the anti-progestagen RU-486 compounds. The results show that the 6 compounds under study were markedly more efficient in terms of growth inhibition in the human T-47D cell line when maintained under anchorage-independent culture conditions than when maintained under anchorage-dependent ones. While RU-486 exhibited a weak statistically significant (p < 0.05) influence on the growth of the T-47D stem cells, tamoxifen had a marked inhibitory influence on the growth of these cells. Of the four retinoids, 4-HPR was the least effective since the lowest doses tested (1 and 0.1 nM) exhibited no statistically (p > 0.05) significant influence on the growth of the stem cells. The most efficient retinoid was TTNPB. It was only at the highest dose (10 microM) that tamoxifen and RU-486 showed a weak inhibitory influence on the growth of the T-47D non-stem cells while all 4 retinoids exerted a significant inhibitory influence on the growth of these non-stem cells, with 4-HPR being the most efficient (P < 0.001) at the highest dose, but ineffective (P > 0.05) at the lowest. Tamoxifen and TTNPB were tested in vivo on hormone-sensitive (HS) and hormone-insensitive (HI) strains of the MXT murine mammary carcinoma. While TTNPB appeared to be equally efficient in terms of growth inhibition in both MXT-HS and MXT-HI models, tamoxifen had only a marginal inhibitory influence on the growth of the MXT-HI strain but did inhibit growth in the case of the MXT-HS one. TTNPB was markedly more efficient than tamoxifen in terms of both inhibiting the cell proliferation level (measured by means of computer-assisted microscopy applied to Feulgen-stained nuclei, a method which enables the percentage of cells in the S phase of the cell cycle to be determined) and triggering cell death (measured by means of the determination of the transglutaminase activity) in both the MXT-HI and MXT-HS models. The very significant TTNPB-induced inhibition of the macroscopic MXT-HS growth rate relates to the triggering of cell death (apoptosis) rather than to an inhibition of cell proliferation. All these results clearly indicate that retinoids are very efficient agents against breast cancer, at least as efficient as tamoxifen.