Simultaneous detection of multiple toxic gases in the air using room temperature gas sensors is significant in low-power environmental monitoring applications. However, the low-temperature resistive gas sensors are sensitive to more than one gas, and thus, an array of gas sensors and high energy-consuming machine learning algorithms are required to predict the concentrations of the individual gases in mixed target gas. Here, we report a computationally less intensive method to predict the composition of the target gases using linear gas sensors. A sensor array consisting of two ZnS resistive gas sensors biased at different voltages in conjunction with the superposition principle is used to predict the concentration of individual gases in the binary mixture of NH3 and CO present in the air. Further, the effect of humidity on response is mitigated by formulating the sensitivity of the sensors as a function of relative humidity. The proposed algorithm predicted the concentration of the individual gases in mixed gas with a maximum absolute error of ∼15% irrespective of humidity levels, which is practically allowed in most gas sensing applications. As the superposition principle is a low-power consuming technique, the proposed approach can be used in applications where trace levels of gases in mixed targets need to be detected with energy-efficient methods. © 2022 The Authors. Published by American Chemical Society.