Denitrifier community phenotypes often result in transient accumulation of denitrification (NO3−→NO2−→NO→N2O→N2) intermediates. Consequently, anoxic spells drive NO-, N2O- and possibly HONO-emissions to the atmosphere, affecting both climate and tropospheric chemistry. Soil pH is a key controller of intermediate levels, and while there is a clear negative correlation between pH and emission of N2O, NO2− concentrations instead increase with pH. These divergent trends are probably a combination of direct effects of pH on the expression/activity of denitrification enzymes, and an indirect effect via altered community composition. This was studied by analyzing metagenomics/transcriptomics and phenomics of two soil denitrifier communities, one of pH 3.8 (Soil3.8) and the other 6.8 (Soil6.8). Soil3.8 had severely delayed N2O reduction despite early transcription of nosZ, encoding N2O reductase, by diverse denitrifiers, and of several nosZ accessory genes. This lends support to a post-transcriptional, pH-dependent mechanism acting on the NosZ apo-protein or on enzymes involved in its maturation. Metagenome/metatranscriptome reads of nosZ were almost exclusively clade I in Soil3.8 while clade II dominated in Soil6.8. Reads of genes and transcripts for NO2−-reductase were dominated by nirK over nirS in both soils, while qPCR-based determinations showed the opposite, demonstrating that standard primer pairs only capture a fraction of the nirK community. The -omics results suggested that low NO2− concentrations in acidic soils, often ascribed to abiotic degradation, are primarily due to enzymatic activity. The NO reductase gene qnor was strongly expressed in Soil3.8, suggesting an important role in controlling NO. Production of HONO, for which some studies claim higher, others lower, emissions from NO2− accumulating soil, was estimated to be ten times higher from Soil3.8 than from Soil6.8. The study extends our understanding of denitrification-driven gas emissions and the diversity of bacteria involved and demonstrates that gene and transcript quantifications cannot always reliably predict community phenotypes.