Extraction efficiency, reagent specificity and selectivity, and element redistribution are potential problems with trace element fractionation by sequential extraction. As part of a larger study of Cd reactivity in soils, we optimized a sequential extraction procedure for accurate, reproducible Cd fractionation using four soil samples and two soil standard reference materials diverse in Cd source, physicochemical properties, and total extractable Cd (CdT, varying from 22 to 42 mg kg−1). Cadmium was partitioned into five operationally defined fractions: 0.1 M Sr(NO3)2 (F1, soluble–exchangeable); 1 M Na acetate, pH 5.0 (F2, sorbed–carbonate); 5% NaOCl, pH 8.5 (F3, oxidizable); 0.4 M oxalate + 0.1 M ascorbate (F4, reducible); and 3 HNO3:1 HCl (F5, residual). By repeating treatments at F1, F3, F4, and F5, we maximized the amount of Cd released for these respective steps. Supernatant pH was used to evaluate carbonate dissolution at F2. Multi-element analyses were used to assess reagent specificity/selectivity. Cd redistribution was estimated by extraction with Pb acetate. Reagent specificity and selectivity were good, suggesting the dissolution of major components at targeted phases (e.g., high Ca in F2). In general, redistribution was minimal (≤3%), but reached 12% for F3 of the sludge-amended soil. Quantitative, reproducible recovery of Cd (96.5 ± 2.1%) was obtained across all samples and averaged 11, 32, 40, 8, and 6% CdT in the respective five fractions. Fractionation trends reflect the Cd sources and physicochemical properties of the samples with Cd being dominant in F3 for soils high in organic matter or contaminated by metal sulfides.