Spinal muscular atrophy (SMA) is the leading genetic cause of death in young children, arising from homozygous deletion or mutation of the survival motor neuron 1 (SMN1) gene. SMN protein expressed from a paralogous gene, SMN2, is the primary genetic modifier of SMA; small changes in overall SMN levels cause dramatic changes in disease severity. Thus, deeper insight into mechanisms that regulate SMN protein stability should lead to better therapeutic outcomes. Here, we show that SMA patient-derived missense mutations in the Drosophila SMN Tudor domain exhibit a pronounced temperature sensitivity that affects organismal viability, larval locomotor function and adult longevity. These disease-related phenotypes are domain specific and result from decreased SMN stability at elevated temperature. This system was utilized to manipulate SMN levels during various stages of Drosophila development. Owing to a large maternal contribution of mRNA and protein, Smn is not expressed zygotically during embryogenesis. Interestingly, we find that only baseline levels of SMN are required during larval stages, whereas high levels of the protein are required during pupation. This previously uncharacterized period of elevated SMN expression, during which the majority of adult tissues are formed and differentiated, could be an important and translationally relevant developmental stage in which to study SMN function. Taken together, these findings illustrate a novel in vivo role for the SMN Tudor domain in maintaining SMN homeostasis and highlight the necessity for high SMN levels at crucial developmental time points that are conserved from Drosophila to humans.