Glutathione S-transferases (GSTs), detoxification enzymes that catalyze the addition of glutathione (GSH) to diverse electrophilic molecules, are often overexpressed in various tumor cells. While fluorescent probes for GSTs have often adopted the 2,4-dinitrobenzenesulfonyl (DNs) group as the receptor unit, they usually suffer from considerable background reaction noise with GSH due to excessive electron deficiency. However, weakening this reactivity is generally accompanied by loss of sensitivity for GSTs, and therefore, finely turning down the reactivity while maintaining certain sensitivity is critical for developing a practical probe. Here, we report a rational semiquantitative strategy for designing such a practical two-photon probe by introducing a parameter adopted from the conceptual density functional theory (CDFT), the local electrophilicity ω k , to characterize this reactivity. As expected, kinetic studies established ω k as efficient to predict the reactivity with GSH, and probe NI3 showing the best performance was successfully applied to detecting GST activities in live cells and tissue sections with high sensitivity and signal-to-noise ratio. Photoinduced electron transfer of naphthalimide-based probes, captured by femtosecond transient absorption for the first time and unraveled by theoretical calculations, also contributes to the negligible background noise.