We are considering the problem of reconstructing indoor temperature distributions using the Acoustic travel-time TOMography (ATOM) technique. Motivated by the benefits of acoustic rooms, we reconstruct the field from the estimated time-of-flight (TOF) of the early reflections, considering the first, second, and third orders. Here, TOF collected from early reflections is used as an input dataset to reconstruct indoor air temperature using the simultaneous iterative reconstruction technique (SIRT). Results showed that even with only one loudspeaker and one microphone, we obtain promising reconstruction fields. However, in order to control the semiconvergence of the SIRT algorithm, several examples of reconstructed fields are provided to highlight the effects of linkage between voxels (so-called weighting factor w ), relaxation parameter ( λ ), and the number of iterations ( M ). Based on the findings, we suggest the optimal values of w , λ , and M , which assist in improving the quality of the reconstructed field. On the other hand, obtaining high-quality indoor field measurements is deeply related to the coordinates of the transceiver, highlighting the need for an optimal position choice. In this context, we propose an improved version of an existing numerical method for predicting the optimal coordinates and reconstructing a highly accurate temperature within an echoic box ( 1.33× 1.0×1.27 m). We show that the predicted optimal position successfully recovers a highly accurate field, satisfying the negative temperature coefficient (NTC)-thermistors measurement.