The 1989–1990 eruption of Redoubt Volcano, Alaska, provided an opportunity to compare petrologic estimates of SO₂ and Cl emissions with estimates of SO₂ emissions based on remote sensing data and estimates of Cl emissions based on plume sampling. In this study, we measure the sulfur and chlorine contents of melt inclusions and matrix glasses in the eruption products to determine petrologic estimates of SO₂ and Cl emissions. We compare the results with emission estimates based on COSPEC and TOMS data for SO₂ and data for Cl/SO₂ in plume samples. For the explosive vent clearing period (December 14–22, 1989), the petrologic estimate for SO₂ emission is 21,000 tons, or ~12% of a TOMS estimate of 175,000 tons. For the dome growth period (December 22, 1989 to mid-June 1990), the petrologic estimate for SO₂ emission is 18,000 tons, or ~3% of COSPEC-based estimates of 572,000–680,000 tons. The petrologic estimates give a total SO₂ emission of only 39,000 tons compared to an integrated TOMS/COSPEC emission estimate of ~1,000,000 tons for the whole eruption, including quiescent degassing after mid-June 1990. Petrologic estimates also appear to underestimate Cl emissions, but apparent HCl scavenging in the plume complicates Cl emission comparisons. Several potential sources of ‘excess sulfur’ often invoked to explain petrologic SO₂ deficits are concluded to be unlikely for the 1989–1990 Redoubt eruption — e.g., breakdown of sulfides, breakdown of anhydrite, release of SO₂ from a hydrothermal system, degassing of commingled infusions of basalt in the magma chamber, and syn-eruptive degassing of sulfur from melt present in non-erupted magma. Leakage and/or diffusion of sulfur from melt inclusions do not provide convincing explanations for the petrologic SO₂ deficits either. The main cause of low petrologic estimates for SO₂ is that melt inclusions do not represent the total sulfur content of the Redoubt magmas, which were vapor-saturated magmas carrying most of their sulfur in an accumulated vapor phase. Almost all the sulfur of the SO₂ emissions was present prior to emission as accumulated magmatic vapor at 6–10 km depth in the magma that supplied the eruption; whole-rock normalized concentrations of gaseous excess S in these magmas remained at ~0.2 wt.% throughout the eruption, equivalent to ~0.7 vol.% at depth. Data for CO₂ emissions during the eruption indicate that CO₂ at whole-rock concentrations of ~0.6 wt.% in the erupted magma was a key factor in creating the vapor saturation and accumulation condition making a vapor phase source of excess sulfur possible at depth. When explosive volcanism involves magma with accumulated vapor, melt inclusions do not provide a sufficient basis for predicting SO₂ emissions. Thus, petrologic estimates made for SO₂ emissions during explosive eruptions of the past may be too low and may significantly underestimate impacts on climate and the chemistry of the atmosphere.