Nanosensors have been widely used for detecting toxic and volatile gases present in indoor and outdoor environment due to their swift response and recovery characteristics. They exhibit high selectivity, sensitivity, quick response and recovery signatures towards target gases. However, determination of the response and recovery time from the transient response of the chemiresistive sensor needs to be accurate to exactly represent the figure of merits of nanosensors. In this context, a non-linear model has been employed to analyse the transient response and recovery curves of a chemiresistive ammonia sensor using nanostructured cobalt-doped-ZnO thin film. Modified Hill equations have been developed based on sigmoidal behaviour of the transient curves. Based on the error analysis, only the response t95% and recovery t95% model showed best results in validation with a low relative prediction error of 0.003 and 0.001 and low root mean square error for cross validation of 9.5 and 4.68 respectively. © 2018 Elsevier Ltd