In previous reports illustrating the effects of conformational restriction of the N-terminal region of human pancreatic growth hormone releasing factor, we demonstrated that D-amino acid substitutions in either of positions 1, 2, or 3 resulted in greatly increased growth hormone releasing activity both in vivo and in vitro. The most active compound, [D-Ala-2]GRF(1-29)NH2, was 51 times more active than the parent 29 amino acid peptide in the sodium pentobarbital anesthetized rat. These observations have now been extended to analogues containing multiple D-amino acid replacements in these three positions. Once again, peptides with superagonist potencies ranging from 1200% to 3800% were obtained after solid-phase synthesis and purification by medium-pressure reverse-phase liquid chromatography. In addition, it was found that [D-Asn-8]- and [D-Ala-4]GRF(1-29)NH2 were, respectively, 2.43 and 1.1 times more active than GRF(1-29)NH2 itself. In contrast, [D-Phe-6] and [D-Thr-7] analogues were virtually inactive. Chou-Fasman structural predictions suggest that the first three residues of the peptide assume no fixed type of conformation but that a reverse turn could be present between residues 6 and 10. Attempts are made to rationalize the biological results with these calculations. The effects of other side chains on the D-amino acid in position 2 were also investigated. Both the Ac-[D-Phe-2]- and Ac-[D-Arg-2]peptides had very low activity. Several of the inactive peptides were tested as possible antagonists of GRF; however, none was able to block the stimulatory effects of GRF(1-29)NH2 after combined administration.