Increased understanding of surface residue decomposition may improve cover crop management in no‐till systems. Most decomposition studies of cover crop residues have been conducted with samples composed of a mixture of leaves and stems. Because leaves and stems have different composition, however, they would be expected to show different mineralization kinetics. The objective of this work was to study C and N mineralization from isolated stems, isolated leaves, and a mix of leaves and stems of wheat (Triticum aestivum L.), rye (Secale cereale L.) oat (Avena sativa L.), and crimson clover (Trifolium incarnatum L.). Cecil loamy sand soil (clayey, kaolinitic, thermic Typic Kanhapludult) was packed into acrylic plastic cylinders, adjusted to 55% water‐filled porosity, treated with either leaves, stems, or a mix of both (1‐cm pieces) on the surface, and incubated at 35°C for 160 d. Air samples for CO2 and N2O determinations were taken periodically and NH3 evolved was trapped during the first 16 d. Soil columns were leached periodically and leachates were analyzed for N (total and inorganic) and total C. The dynamics of C and N mineralization of a mix of leaves and stems was different from the patterns predicted from isolated leaves and isolated stems. These results indicate a strong interaction between stems and leaves during early stages of decomposition, which may be relevant for predicting N mineralization from cover crop residues. The best predictors for N mineralization were residue C/N ratio and the reciprocal of residue N concentration.