Fluid inclusions are the only available samples of paleo-fluids responsible for crystallization of hydrothermal minerals including ore phases. Analysis of fluid inclusions implicitly assumes that the inclusions have preserved their chemical composition since the time of their entrapment. There is, however, an increasing evidence from experimental work and analytical studies of natural samples showing that inclusions hosted in quartz – a ubiquitous host in many ore-forming systems – can experience post-entrapment modification. The contents of small monovalent cations, e.g. Cu, can be enhanced up to several orders of magnitude in vapor-dominated fluid inclusions by diffusional exchange for hydrogen. In this study, I focus on chemical analysis of fluid inclusions hosted by various minerals that co-precipitated with quartz and host the same fluid, in order to compare their compositional consistency and address existence and extent of post-entrapment modification. The fluid inclusions were analyzed by laser ablation-inductively coupled plasma mass spectrometry (LA-ICP-MS) and I explore analytical possibilities and challenges when applied to analysis of ultra-trace quantities of elements in small amounts of fluid, such as Au contents of magmatic vapor. I use representative samples from mineralization settings at convergent plate boundaries, namely granite-related Sn-W deposit (Mole Granite, Eastern Australia) and porphyry Cu deposit (Bingham Canyon, Utah, USA). In these systems, highly saline brines are important carriers of chloride-complexed metals and coexist with sulfur-rich vapors containing sulfide-complexing metals, e.g. As, Au, and, arguably, to various degree Cu. In the Mole Granite Sn-W deposit in Eastern Australia brine and vapor coexist in hydrothermal quartz, topaz, and beryl. Fluid inclusions in these three minerals show consistent salinities of 28–39 wt. % NaCleq in brine and 1–6 wt. % NaCleq in vapor at formation temperatures above 380 °C. These two fluid types have nearly identical concentrations of trace elements such as B, Na, K, S, and Cs, but reveal significant discrepancies in Cu abundances. The Cu concentrations in the quartz-hosted vapor inclusions are in the range of 5–50 000 ppm, in contrast to 20–2000 ppm Cu found in topaz- and beryl-hosted vapor inclusions. Partition coefficients of Cu between brine and vapor recorded by the topaz- and beryl-hosted fluids are in agreement with experimental results for magmatic-hydrothermal conditions, and in favor of the brine. The quartz-hosted fluids, however, show partitioning coefficients of Cu that exceed those found in all available experiments in sulfur-bearing systems. The vapor inclusions preserved in quartz are thus interpreted to have undergone significant post-entrapment modification, facilitated by their initially high sulfur content. By contrast, Au in the brine and vapor inclusions has been detected as erratic values only, despite that the detection limits of Au achieved by the LA-ICP-MS analysis were as low as single ppb. This extreme depletion in Au has been addressed by thermodynamic modeling indicating that saturation of the fluid with liquid bismuth can cause effective and essentially complete removal of Au from the fluid, and precipitates both metals as Bi-Au alloy. In the Bingham Canyon Cu-Mo-Au deposit (Utah, USA), garnet, pyroxene and quartz coexist in carbonate-replacement skarns that trapped brine, vapor as well as initial single-phase fluid prior to its separation. The single-phase intermediate-density fluid with an average salinity of 15 wt. NaCleq was trapped above approx. 550 °C and 800 bar. Salinities of brine and vapor inclusions are 32–42 and 2–7 wt. % NaCleq, respectively, and they formed at 420–380 °C and approx. 200 bar. The characteristics of corresponding fluid types in garnet, quartz, and to some degree in pyroxene remain nearly identical including their metal and other trace element contents. With improved analytical configuration, reliable Au signals were obtained from the fluid inclusions. Both Cu and Au have approximately equal content relative to Na in vapor and brine. This contrasts with previous fluid analyses from the porphyry quartz stockwork suggesting that majority of Cu and Au was transported by vapor. The new data suggest that the Cu concentrations in the vapor from the quartz stockwork were likely enhanced after inclusion entrapment, while the Au concentrations in the vapor phase require verification using lower instrumental detection limits. The Cu, Au and S concentration in the fluids from the porphyry quartz stockwork are consistently higher (0.2–5 ppm Au, 0.05–3 wt. % Cu) than those from the skarn minerals (0.02–0.4 ppm Au, 0.006–0.3 wt. % Cu), in contrast to ore grades that vary in an opposite manner (Au/Cu of 10-5 in the porphyry vs. 5∙10-5 in the skarn). This dichotomy was likely caused by chemical reactivity of carbonate precursor to the skarns and fluid focusing promoted by porosity and permeability enhancement during its decarbonation. In broader context, post-entrapment modification of vapor inclusions appears to be related to acidic and sulfur-rich fluids such as those preserved in granite- or porphyry-hosted pathways. By contrast, skarn-forming reactions appear to effectively neutralize acidity and together with low sulfur concentrations, the skarn-hosted fluids lack the principal driving forces that allow chemical-potential coupling and metal diffusion between fluid and host mineral, hence control the extent of post-entrapment modification. This study illustrates novel approaches to analysis of ultra-low metal concentrations in small volumes of fluids and analysis of fluid inclusions in non-traditional silicate minerals. I have identified and addressed several outstanding challenges of the LA-ICP-MS analytics, specifically effects of contamination on reliable analysis of elements close to their detection limits, strategies for reduction of the analytical data, and estimation of average concentrations for fluid inclusion populations.