We investigated plagioclase phenocrysts in dacite of the 2004-5 eruption of Mount St. Helens to gain insights into the magmatic processes of the current eruption, which is characterized by prolonged, nearly solid-state extrusion, low gas emission, and shallow seismicity. In addition, we investigated plagioclase of 1980-86 dacite. Light and Nomarski microscopy were used to texturally characterize plagioclase crystals. Electron microprobe analyses measured their compositions. We systematically mapped and categorized all plagioclase phenocrysts in a preselected area according to the following criteria: (1) occurrence of zones of acicular orthopyroxene inclusions, (2) presence of dissolution surface(s), and (3) spatial association of 1 and 2. Phenocrysts fall into three main categories; one category contains four subcategories. The range of anorthite (An) content in 2004-5 plagioclase is about An_57-35 during the last 30-40 percent crystallization of plagioclase phenocrysts. Select microphenocrysts (10-50 μm) range from An₃₀ to An₄₂. Anorthite content is lowest near outermost rims of phenocrysts, but zonation patterns between interior and rim indicate variable trends that correlate with textural features. Crystals without dissolution surfaces (about 14 percent of total) show steadily decreasing An content outward to the crystal rim (outer ~80 μm). All other crystals are banded as a consequence of dissolution; dissolution surfaces are band boundaries. Such crystals display normal outward An zoning within a single band that, following dissolution, is then overgrown abruptly by high-An material of the next band. Swarms of acicular orthopyroxene inclusions in plagioclase are characteristic of 2004-5 dacite. They occur mostly inward of dissolution surfaces, where band composition reaches lowest An content. The relative proportions of the three crystal types are distinctly different between 2004-5 dacite and 1980s dome dacite. We propose that crystals with no dissolution surfaces are those that were supplied last to the shallow reservoir, whereas plagioclase with increasingly more complex zoning patterns (that is, the number of zoned bands bounded by dissolution surfaces) result from prolonged residency and evolution in the reservoir. We propose that banding and An zoning across multiple bands are primarily a response to thermally induced fluctuations in crystallinity of the magma in combination with recharge; a lesser role is ascribed to cycling crystals through pressure gradients. Crystals without dissolution surfaces, in contrast, could have grown only in response to steady(?) decompression. Some heating-cooling cycles probably postdate the final eruption in 1986. They resulted from small recharge events that supplied new crystals that then experienced resorption-growth cycles. We suggest that magmatic events shortly prior to the current eruption, recorded in the outermost zones of plagioclase phenocrysts, began with the incorporation of acicular orthopyroxene, followed by last resorption, and concluded with crystallization of euhedral rims. Finally, we propose that 2004-5 dacite is composed mostly of dacite magma that remained after 1986 and underwent subsequent magmatic evolution but, more importantly, contains a component of new dacite from deeper in the magmatic system, which may have triggered the new eruption.