This thesis details my investigations of marine cyanobacterial natural products that resulted in the discovery of eighteen new secondary metabolites. Isolation and characterization of these unique molecules were carried out using different chromatographic techniques and by careful analyses of 1D and 2D NMR data, respectively. Preliminary bioassays of the crude organic extract of the marine cyanobacterium Lyngbya semiplena collected from Papua New Guinea in 1999, showed good activity in the brine shrimp toxicity model at 10 ppm. Guided by this assay, seven new anandamide-like fatty acid amides, semiplenamides A to G, together with four cyclic depsipeptides, wewakpeptins A to D, were identified. Due to the structural resemblance of the novel ethanolamide derivatives (semiplenamide A-G) with anandamide, (N-arachidonoyl-ethanolamine), an endogenous agonist of cannabinoid receptors compounds, semiplenamide A-G were tested on the well characterized proteins of the endocannabinoid system: 1) the "central" cannabinoid CB₁ receptors; 2) the anandamide membrane transporter (AMT), which is responsible for anandamide cellular uptake; and 3) the fatty acid amide hydrolase (FAAH), which catalyses anandamide hydrolysis. Three showed modest potency in displacing radiolabeled anandamide from the cannabinoid receptor (CB₁), and one was a modest inhibitor of the anandamide membrane transporter (AMT). The wewakpeptins were tested for cytotoxicity to NCI-H460 human lung tumor and neuro-2a mouse neuroblastoma cells. Intriguingly, wewakpeptin A and B were approximately 10-fold more toxic than C and D to these cell lines. Lyngbya majuscula has been recognized as a chemically and biologically rich strain. Five new lyngbyabellin analogs, lyngbyabellins E-I, along with the known compound dolabellin, originally isolated from sea hare Dolabella auricularia, were identified from a 2002 Papua New Guinea collection of the marine cyanobacterium Lyngbya majuscula. The lyngbyabellins were tested for cytotoxicity to NCI-H460 human lung tumor and neuro-2a mouse neuroblastoma cells and had LC₅₀ values between 0.2 and 4.8 μM. Intriguingly, lyngbyabellin E and H appeared to be more active against the H460 cell line with LC₅₀ values of 0.4 μM and 0.2 μM, respectively, compared to LC₅₀ values of 1.2 and 1.4 μM in the neuro-2a cell line. Lynbyabellin I was the most toxic to neuro-2a cells (LC₅₀ 0.7 μM), whereas lyngbyabellin G, was the least cytotoxic of all compounds to either cell line. On the basis of this limited screening, it appears that lung tumor cell toxicity is enhanced in the cyclic representatives, and overall potency is increased in those containing an elaborated side chain. Additionally, two new cytotoxins, aurilides B and C, which are closely related to aurilide, originally isolated from the sea hare Dolabella auricularia, have been identified from the same extract where the lyngbyabellins E-I were isolated. Aurilides B and C were tested for cytotoxicity to NCI-H460 human lung tumor and neuro-2a mouse neuroblastoma cells. Interestingly, aurilide B was approximately 4-fold more toxic than C to these cell lines. The LC₅₀ for Aurilide B was 0.01 μM and 0.04 μM for neuro-2a and H460 cells, respectively, and 0.05 μM and 0.13 μM, respectively, for aurilide C. Aurilide B was evaluated in the NCI 60 cell line panel and found to exhibit a high level of growth inhibition in leukemia, renal, and prostate cancer cell lines with a GI₅₀ less than 10 nM. Aurilide B showed net tumor cell killing activity in the NCI's hollow fiber assay, an in vivo model for assessing a chemical's anticancer activity. / Graduation date: 2006