Ru-catalyzed olefin metathesis is an exceptionally powerful, versatile methodology for the assembly of carbon–carbon bonds. The N-heterocyclic carbene (NHC)-stabilized, “second-generation” Ru catalysts have enabled groundbreaking recent advances, ranging from the RCM assembly of cyclic peptides as hepatitis C virus therapeutics, to the elaboration of renewable seed oils and phenylpropanoids into value-added products and chemicals. However, key limitations arise from facile catalyst decomposition. Despite a plethora of studies on the synthesis of new catalysts, and on the decomposition processes accessible to the precatalyst and resting-state species, the underlying principles that govern decomposition of the active intermediates have been surprisingly little examined. One important reason for this is their incredible reactivity: the four-coordinate methylidene intermediate RuCl2(H2IMes)(=CH2) is too short-lived to be observed, while the metallacyclobutane (MCB) intermediate RuCl2(H2IMes)(2-C3H6) can only be observed below –40 °C. This makes them extremely challenging, but also fascinating targets for study. Understanding the underlying chemistry that dictates their reactivity and decomposition is essential for informed catalyst and process redesign, and is thus of fundamental interest, but also considerable practical importance. This thesis work thus aims at understanding the decomposition of active intermediates relevant to the highly-active, second-generation class of catalysts. Emphasis is placed on examining a variety of metathesis contexts, as well as providing solutions. Treated first are the decomposition pathways that arise during metathesis of electron-deficient olefins, a frontier area in organic synthesis, and in the utilization of renewable resources. An unexpected correlation is revealed between rapid catalyst decomposition, and the presence of a stabilizing PCy3 ligand in the standard catalyst for this reaction. The nucleophilic phosphine ligand is shown to attack an acrylate olefin, forming enolates that function as potent Brønsted bases. Literature evidence suggests that such strong bases are innocuous towards the precatalyst, pointing towards a key role for the active intermediates in Brønsted base-induced catalyst decomposition. Precisely which intermediate is involved, as well as the site of deprotonation, is elucidated next. Prior to this work, the NHC ligand was widely believed to be the target for attack. However, through labelling experiments, analysis of the Ru and organic byproducts, and computational studies, deprotonation is shown to occur at the MCB ring. Moreover, MCB deprotonation is revealed to be unexpectedly general, and not contingent on the presence of either an exceptionally strong base, or an electron-deficient substrate. This understanding is key, given recent reports from pharma highlighting the adverse impact of base contaminants, as well as current interest in metathesis of amine-containing substrates. Next examined are the intrinsic decomposition pathways operative for the MCB and four-coordinate methylidene. Prior to this work, the only reported pathway for decomposition of these two species involved beta-elimination of the MCB ring as propene. However, beta-elimination is shown to play an unexpectedly minor role in catalyst decomposition: less than 40% propenes are observed, even under conditions expected to favour MCB elimination. Bimolecular coupling of the methylidene, with loss of the methylidene moiety as ethylene, is proposed to account for the difference. Thus, transiently-stabilized adducts RuCl2(H2IMes)(=CH2)(L)n (L = o-dianiline or pyridine) are synthesized at temperatures down to –120 °C. On warming, these adducts lose Ln and rapidly decompose via bimolecular coupling, with loss of the methylidene moiety as ethylene. These experiments provide the first unambiguous evidence for bimolecular coupling in the important "second-generation" Ru systems, nearly two decades after which this pathway was dismissed in leading papers and reviews. The last two sections focus on solutions. First, a powerful, straightforward solution to the “enolate problem” is developed, whereby the acrylate enolates are quenched and sequestered via reaction with a polyphenol resin. Then, methods for preventing catalyst decomposition during matrix-assisted laser desorption / ionization mass spectrometry (MALDI-MS) are developed, via elucidation of the instrumental and experimental factors that promote successful analysis. As one of the only MS methods capable of affording insight into neutral metal complexes and catalysts, MALDI has unique potential to enable routine analysis of catalyst speciation and decomposition in situ, under real catalytic conditions, for a wide range of catalytic reactions. Collectively, the findings in this thesis offer a much more complete understanding of the fundamental pathways accessible to the important, highly-active metathesis intermediates, and offer strategies likely to inform practice in both academic and industrial settings. This understanding is key to harnessing the full potential of metathesis methodologies.