Reactive oxygen species (ROS), including superoxide anion and hydrogen peroxide (H2O2), have a diverse array of physiological and pathological effects within living cells depending on the extent, timing, and location of their production. For measuring ROS production in cells, the ESR spin trapping technique using cyclic nitrones distinguishes itself from other methods by its specificity for superoxide and hydroxyl radical. However, several drawbacks, such as the low spin trapping rate and the spontaneous and cell-enhanced decomposition of the spin adducts to ESR-silent products, limit the application of this method to biological systems. Recently, new cyclic nitrones bearing a triphenylphosphonium (Mito-DIPPMPO) or a permethylated β-cyclodextrin moiety (CD-DIPPMPO) have been synthesized and their spin adducts demonstrated increased stability in buffer. In this study, a comparison of the spin trapping efficiency of these new compounds with commonly used cyclic nitrone spin traps, i.e., 5,5-dimethyl-1-pyrroline N-oxide (DMPO), and analogs BMPO, DEPMPO, and DIPPMPO, was performed on RAW 264.7 macrophages stimulated with phorbol 12-myristate 13-acetate. Our results show that Mito-DIPPMPO and CD-DIPPMPO enable a higher detection of superoxide adduct, with a low (if any) amount of hydroxyl adduct. CD-DIPPMPO, especially, appears to be a superior spin trap for extracellular superoxide detection in living macrophages, allowing measurement of superoxide production in unstimulated cells for the first time. The main rationale put forward for this extreme sensitivity is that the extracellular localization of the spin trap prevents the reduction of the spin adducts by ascorbic acid and glutathione within cells.