A 34-amino acid peptide has been synthesized based on an amino acid sequence from the third domain of native full-length alpha-fetoprotein, which has been shown to have both antigrowth and anticancer activities. This peptide, known as the growth-inhibitory peptide (GIP), has two cysteine residues and demonstrates reduced antigrowth activity after long-term storage, presumably due to disulfide bond formation. The disulfide bridge problem was addressed by replacing the two naturally occurring cysteines with either glycines, alanines, or serines (to produce the G-, A- and S-peptides, respectively). The non-hydrophobic G- and S-peptides were found to exist as dimers, while the more hydrophobic C- and A-peptides formed trimers in solution under certain conditions of pH and peptide concentration. The A-peptide was already known to display anticancer activity; however, the G- and S-serine analogs have not been studied in depth since they had demonstrated low antigrowth activities in rodent uterine assays. Using both in vivo and in vitro assays, the A-, G- and S-peptides were shown to exhibit various degrees of cancer growth suppression. An in vitro culture assay, using MCF-7 breast cancer cells, demonstrated that both the G- and S-peptides showed modest cancer growth suppression, while the A- analog showed strong inhibition at doses ranging from 10(-5) M to 10(-7) M. In contrast, an in vivo ascites tumor study of all four peptides showed them to have notable activity in the suppression of mouse mammary tumor growth. Overall, our data indicated that physicochemical properties, such as hydrophobicity, oligomeric state and secondary structure, contribute to the anticancer activity of both the active C- peptide and its analogs. In addition, the antigrowth rodent uterine assay was not always predictive of the anticancer potential of the peptide forms, suggesting a difference between the mechanism of peptide action in the antigrowth models and that in the anticancer assay. Notably, the antigrowth assay failed to predict the marked anticancer activity of the analogs against a mammary tumor, indicating that the growth bioassay cannot always be relied upon as a screening protocol.