Respiratory motion can blur the tomographic reconstruction of PET or SPECT images, which subsequently impair quantitative measurements, e.g. in the upper abdomen area. Respiratory signal phase-based gated reconstruction addresses this problem, but deteriorates the signal-to-noise ratio and other intensity-based quality measures. This article proposes a 3D reconstruction method dedicated to micro-SPECT imaging of mice. From a 4D acquisition, the phase images exhibiting motion are identified and the associated list-mode data are discarded, which enables the reconstruction of a 3D image without respiratory artefacts. The proposed method allows a motion-free reconstruction exhibiting both satisfactory count statistics and accuracy of measures. With respect to standard 3D reconstruction (NG3D) without breathing motion correction, an increase of 14.6% of the mean SUV has been observed, while, with respect to a gated 4D reconstruction (G4D), up to 60% less noise and an increase of up to 124% of the SNR have been demonstrated.