Somatosensory cortex hyperexcitability is present in the pre-symptomatic stage of amyotrophic lateral sclerosis (ALS) as evidenced by brain recordings, but its synaptic basis remains unclear. We examined synaptic plasticity, the density of asymmetric (putative excitatory) and symmetric (putative inhibitory) synapses, dendritic spine morphology, and the putative excitatory/inhibitory (E/I) ratio in the B2 barrel of the somatosensory cortex in female mice of an ALS mouse model. Transgenic mice, B6SJL-Tg (SOD1*G93A)1Gur/J, were used as the ALS model, and wild-type (WT) B6SJL/F1 mice served as controls. ALS mice were allocated to experimental groups based on disease stage (pre-symptomatic, onset, or terminal) and training condition (swim-trained or untrained). Swim training was applied after the first onset of symptoms (clinical score 1). We analyzed and quantified the density of asymmetric (putative excitatory) and symmetric (putative inhibitory) synapses and E/I ratios using serial electron micrographs to understand how these parameters change during disease progression and whether swim training influences this process. Our results showed stage-dependent alterations in asymmetric (putative excitatory) and symmetric (putative inhibitory) synaptic architecture in ALS. The obtained data showed an increase in the excitatory synaptic density in the presymptomatic ALS mice. This finding is consistent with previous reports of early cortical hyperexcitability and may reflect structural alterations associated with an initial increase in excitatory synapses before disease onset. Importantly, we report here an increase in inhibitory synapses at disease onset. TEM-based synaptic density quantification revealed reduced excitatory synapse density in the B2 barrel of the somatosensory cortex of trained ALS mice compared to WT controls, alongside a trend toward a reduced putative excitatory/inhibitory synaptic ratio. However, as no significant differences were detected between trained and untrained ALS mice, the contribution of swim training to these alterations remains unclear. Notably, swim training was not associated with detectable adverse effects on somatosensory cortex ultrastructure, excitatory synapse density, or the putative excitatory/inhibitory ratio, supporting previous observations that swim training is well tolerated under these experimental conditions. To our knowledge, these results provide the first TEM-based ultrastructural characterization of synaptic architecture in swim-trained SOD1-G93A mice, although further studies are needed to establish the underlying mechanisms and therapeutic relevance in ALS.